case study education examples

Center for Teaching

Case studies.

Print Version

Case studies are stories that are used as a teaching tool to show the application of a theory or concept to real situations. Dependent on the goal they are meant to fulfill, cases can be fact-driven and deductive where there is a correct answer, or they can be context driven where multiple solutions are possible. Various disciplines have employed case studies, including humanities, social sciences, sciences, engineering, law, business, and medicine. Good cases generally have the following features: they tell a good story, are recent, include dialogue, create empathy with the main characters, are relevant to the reader, serve a teaching function, require a dilemma to be solved, and have generality.

Instructors can create their own cases or can find cases that already exist. The following are some things to keep in mind when creating a case:

• What do you want students to learn from the discussion of the case?
• What do they already know that applies to the case?
• What are the issues that may be raised in discussion?
• How will the case and discussion be introduced?
• What preparation is expected of students? (Do they need to read the case ahead of time? Do research? Write anything?)
• What directions do you need to provide students regarding what they are supposed to do and accomplish?
• Do you need to divide students into groups or will they discuss as the whole class?
• Are you going to use role-playing or facilitators or record keepers? If so, how?
• What are the opening questions?
• How much time is needed for students to discuss the case?
• What concepts are to be applied/extracted during the discussion?
• How will you evaluate students?

To find other cases that already exist, try the following websites:

• The National Center for Case Study Teaching in Science , University of Buffalo. SUNY-Buffalo maintains this set of links to other case studies on the web in disciplines ranging from engineering and ethics to sociology and business
• A Journal of Teaching Cases in Public Administration and Public Policy , University of Washington

For more information:

• World Association for Case Method Research and Application

Book Review :  Teaching and the Case Method , 3rd ed., vols. 1 and 2, by Louis Barnes, C. Roland (Chris) Christensen, and Abby Hansen. Harvard Business School Press, 1994; 333 pp. (vol 1), 412 pp. (vol 2).

Teaching Guides

• Online Course Development Resources
• Principles & Frameworks
• Pedagogies & Strategies
• Reflecting & Assessing
• Challenges & Opportunities
• Populations & Contexts

Quick Links

• Services for Departments and Schools
• Examples of Online Instructional Modules

• Our Mission

Making Learning Relevant With Case Studies

The open-ended problems presented in case studies give students work that feels connected to their lives.

To prepare students for jobs that haven’t been created yet, we need to teach them how to be great problem solvers so that they’ll be ready for anything. One way to do this is by teaching content and skills using real-world case studies, a learning model that’s focused on reflection during the problem-solving process. It’s similar to project-based learning, but PBL is more focused on students creating a product.

Case studies have been used for years by businesses, law and medical schools, physicians on rounds, and artists critiquing work. Like other forms of problem-based learning, case studies can be accessible for every age group, both in one subject and in interdisciplinary work.

You can get started with case studies by tackling relatable questions like these with your students:

• How can we limit food waste in the cafeteria?
• How can we get our school to recycle and compost waste? (Or, if you want to be more complex, how can our school reduce its carbon footprint?)
• How can we improve school attendance?
• How can we reduce the number of people who get sick at school during cold and flu season?

Addressing questions like these leads students to identify topics they need to learn more about. In researching the first question, for example, students may see that they need to research food chains and nutrition. Students often ask, reasonably, why they need to learn something, or when they’ll use their knowledge in the future. Learning is most successful for students when the content and skills they’re studying are relevant, and case studies offer one way to create that sense of relevance.

Teaching With Case Studies

Ultimately, a case study is simply an interesting problem with many correct answers. What does case study work look like in classrooms? Teachers generally start by having students read the case or watch a video that summarizes the case. Students then work in small groups or individually to solve the case study. Teachers set milestones defining what students should accomplish to help them manage their time.

During the case study learning process, student assessment of learning should be focused on reflection. Arthur L. Costa and Bena Kallick’s Learning and Leading With Habits of Mind gives several examples of what this reflection can look like in a classroom: 

Journaling: At the end of each work period, have students write an entry summarizing what they worked on, what worked well, what didn’t, and why. Sentence starters and clear rubrics or guidelines will help students be successful. At the end of a case study project, as Costa and Kallick write, it’s helpful to have students “select significant learnings, envision how they could apply these learnings to future situations, and commit to an action plan to consciously modify their behaviors.”

Interviews: While working on a case study, students can interview each other about their progress and learning. Teachers can interview students individually or in small groups to assess their learning process and their progress.

Student discussion: Discussions can be unstructured—students can talk about what they worked on that day in a think-pair-share or as a full class—or structured, using Socratic seminars or fishbowl discussions. If your class is tackling a case study in small groups, create a second set of small groups with a representative from each of the case study groups so that the groups can share their learning.

4 Tips for Setting Up a Case Study

1. Identify a problem to investigate: This should be something accessible and relevant to students’ lives. The problem should also be challenging and complex enough to yield multiple solutions with many layers.

2. Give context: Think of this step as a movie preview or book summary. Hook the learners to help them understand just enough about the problem to want to learn more.

3. Have a clear rubric: Giving structure to your definition of quality group work and products will lead to stronger end products. You may be able to have your learners help build these definitions.

4. Provide structures for presenting solutions: The amount of scaffolding you build in depends on your students’ skill level and development. A case study product can be something like several pieces of evidence of students collaborating to solve the case study, and ultimately presenting their solution with a detailed slide deck or an essay—you can scaffold this by providing specified headings for the sections of the essay.

Problem-Based Teaching Resources

There are many high-quality, peer-reviewed resources that are open source and easily accessible online.

• The National Center for Case Study Teaching in Science at the University at Buffalo built an online collection of more than 800 cases that cover topics ranging from biochemistry to economics. There are resources for middle and high school students.
• Models of Excellence , a project maintained by EL Education and the Harvard Graduate School of Education, has examples of great problem- and project-based tasks—and corresponding exemplary student work—for grades pre-K to 12.
• The Interdisciplinary Journal of Problem-Based Learning at Purdue University is an open-source journal that publishes examples of problem-based learning in K–12 and post-secondary classrooms.
• The Tech Edvocate has a list of websites and tools related to problem-based learning.

In their book Problems as Possibilities , Linda Torp and Sara Sage write that at the elementary school level, students particularly appreciate how they feel that they are taken seriously when solving case studies. At the middle school level, “researchers stress the importance of relating middle school curriculum to issues of student concern and interest.” And high schoolers, they write, find the case study method “beneficial in preparing them for their future.”

Using Case Studies to Teach

Why Use Cases?

Many students are more inductive than deductive reasoners, which means that they learn better from examples than from logical development starting with basic principles. The use of case studies can therefore be a very effective classroom technique.

Case studies are have long been used in business schools, law schools, medical schools and the social sciences, but they can be used in any discipline when instructors want students to explore how what they have learned applies to real world situations. Cases come in many formats, from a simple “What would you do in this situation?” question to a detailed description of a situation with accompanying data to analyze. Whether to use a simple scenario-type case or a complex detailed one depends on your course objectives.

Most case assignments require students to answer an open-ended question or develop a solution to an open-ended problem with multiple potential solutions. Requirements can range from a one-paragraph answer to a fully developed group action plan, proposal or decision.

Common Case Elements

Most “full-blown” cases have these common elements:

• A decision-maker who is grappling with some question or problem that needs to be solved.
• A description of the problem’s context (a law, an industry, a family).
• Supporting data, which can range from data tables to links to URLs, quoted statements or testimony, supporting documents, images, video, or audio.

Case assignments can be done individually or in teams so that the students can brainstorm solutions and share the work load.

The following discussion of this topic incorporates material presented by Robb Dixon of the School of Management and Rob Schadt of the School of Public Health at CEIT workshops. Professor Dixon also provided some written comments that the discussion incorporates.

Advantages to the use of case studies in class

A major advantage of teaching with case studies is that the students are actively engaged in figuring out the principles by abstracting from the examples. This develops their skills in:

• Problem solving
• Analytical tools, quantitative and/or qualitative, depending on the case
• Decision making in complex situations
• Coping with ambiguities

Guidelines for using case studies in class

In the most straightforward application, the presentation of the case study establishes a framework for analysis. It is helpful if the statement of the case provides enough information for the students to figure out solutions and then to identify how to apply those solutions in other similar situations. Instructors may choose to use several cases so that students can identify both the similarities and differences among the cases.

Depending on the course objectives, the instructor may encourage students to follow a systematic approach to their analysis.  For example:

• What is the issue?
• What is the goal of the analysis?
• What is the context of the problem?
• What key facts should be considered?
• What alternatives are available to the decision-maker?
• What would you recommend — and why?

An innovative approach to case analysis might be to have students  role-play the part of the people involved in the case. This not only actively engages students, but forces them to really understand the perspectives of the case characters. Videos or even field trips showing the venue in which the case is situated can help students to visualize the situation that they need to analyze.

Accompanying Readings

Case studies can be especially effective if they are paired with a reading assignment that introduces or explains a concept or analytical method that applies to the case. The amount of emphasis placed on the use of the reading during the case discussion depends on the complexity of the concept or method. If it is straightforward, the focus of the discussion can be placed on the use of the analytical results. If the method is more complex, the instructor may need to walk students through its application and the interpretation of the results.

Leading the Case Discussion and Evaluating Performance

Decision cases are more interesting than descriptive ones. In order to start the discussion in class, the instructor can start with an easy, noncontroversial question that all the students should be able to answer readily. However, some of the best case discussions start by forcing the students to take a stand. Some instructors will ask a student to do a formal “open” of the case, outlining his or her entire analysis.  Others may choose to guide discussion with questions that move students from problem identification to solutions.  A skilled instructor steers questions and discussion to keep the class on track and moving at a reasonable pace.

In order to motivate the students to complete the assignment before class as well as to stimulate attentiveness during the class, the instructor should grade the participation—quantity and especially quality—during the discussion of the case. This might be a simple check, check-plus, check-minus or zero. The instructor should involve as many students as possible. In order to engage all the students, the instructor can divide them into groups, give each group several minutes to discuss how to answer a question related to the case, and then ask a randomly selected person in each group to present the group’s answer and reasoning. Random selection can be accomplished through rolling of dice, shuffled index cards, each with one student’s name, a spinning wheel, etc.

Tips on the Penn State U. website: http://tlt.its.psu.edu/suggestions/cases/

If you are interested in using this technique in a science course, there is a good website on use of case studies in the sciences at the University of Buffalo.

Dunne, D. and Brooks, K. (2004) Teaching with Cases (Halifax, NS: Society for Teaching and Learning in Higher Education), ISBN 0-7703-8924-4 (Can be ordered at http://www.bookstore.uwo.ca/ at a cost of $15.00)

• Subject List
• Take a Tour
• For Authors
• Subscriber Services
• Publications
• African American Studies
• African Studies
• American Literature
• Anthropology
• Architecture Planning and Preservation
• Art History
• Atlantic History
• Biblical Studies
• British and Irish Literature
• Childhood Studies
• Chinese Studies
• Cinema and Media Studies
• Communication
• Criminology
• Environmental Science
• Evolutionary Biology
• International Law
• International Relations
• Islamic Studies
• Jewish Studies
• Latin American Studies
• Latino Studies
• Linguistics
• Literary and Critical Theory
• Medieval Studies
• Military History
• Political Science
• Public Health
• Renaissance and Reformation
• Social Work
• Urban Studies
• Victorian Literature
• Browse All Subjects

How to Subscribe

• Free Trials

In This Article Expand or collapse the "in this article" section Case Study in Education Research

Introduction, general overview and foundational texts of the late 20th century.

• Conceptualisations and Definitions of Case Study
• Case Study and Theoretical Grounding
• Choosing Cases
• Methodology, Method, Genre, or Approach
• Case Study: Quality and Generalizability
• Multiple Case Studies
• Exemplary Case Studies and Example Case Studies
• Criticism, Defense, and Debate around Case Study

Related Articles Expand or collapse the "related articles" section about

About related articles close popup.

Lorem Ipsum Sit Dolor Amet

Vestibulum ante ipsum primis in faucibus orci luctus et ultrices posuere cubilia Curae; Aliquam ligula odio, euismod ut aliquam et, vestibulum nec risus. Nulla viverra, arcu et iaculis consequat, justo diam ornare tellus, semper ultrices tellus nunc eu tellus.

• Data Collection in Educational Research
• Mixed Methods Research
• Program Evaluation

Other Subject Areas

Forthcoming articles expand or collapse the "forthcoming articles" section.

• English as an International Language for Academic Publishing
• Girls' Education in the Developing World
• History of Education in Europe
• Find more forthcoming articles...
• Export Citations
• Share This Facebook LinkedIn Twitter

Case Study in Education Research by Lorna Hamilton LAST REVIEWED: 27 June 2018 LAST MODIFIED: 27 June 2018 DOI: 10.1093/obo/9780199756810-0201

It is important to distinguish between case study as a teaching methodology and case study as an approach, genre, or method in educational research. The use of case study as teaching method highlights the ways in which the essential qualities of the case—richness of real-world data and lived experiences—can help learners gain insights into a different world and can bring learning to life. The use of case study in this way has been around for about a hundred years or more. Case study use in educational research, meanwhile, emerged particularly strongly in the 1970s and 1980s in the United Kingdom and the United States as a means of harnessing the richness and depth of understanding of individuals, groups, and institutions; their beliefs and perceptions; their interactions; and their challenges and issues. Writers, such as Lawrence Stenhouse, advocated the use of case study as a form that teacher-researchers could use as they focused on the richness and intensity of their own practices. In addition, academic writers and postgraduate students embraced case study as a means of providing structure and depth to educational projects. However, as educational research has developed, so has debate on the quality and usefulness of case study as well as the problems surrounding the lack of generalizability when dealing with single or even multiple cases. The question of how to define and support case study work has formed the basis for innumerable books and discursive articles, starting with Robert Yin’s original book on case study ( Yin 1984 , cited under General Overview and Foundational Texts of the Late 20th Century ) to the myriad authors who attempt to bring something new to the realm of case study in educational research in the 21st century.

This section briefly considers the ways in which case study research has developed over the last forty to fifty years in educational research usage and reflects on whether the field has finally come of age, respected by creators and consumers of research. Case study has its roots in anthropological studies in which a strong ethnographic approach to the study of peoples and culture encouraged researchers to identify and investigate key individuals and groups by trying to understand the lived world of such people from their points of view. Although ethnography has emphasized the role of researcher as immersive and engaged with the lived world of participants via participant observation, evolving approaches to case study in education has been about the richness and depth of understanding that can be gained through involvement in the case by drawing on diverse perspectives and diverse forms of data collection. Embracing case study as a means of entering these lived worlds in educational research projects, was encouraged in the 1970s and 1980s by researchers, such as Lawrence Stenhouse, who provided a helpful impetus for case study work in education ( Stenhouse 1980 ). Stenhouse wrestled with the use of case study as ethnography because ethnographers traditionally had been unfamiliar with the peoples they were investigating, whereas educational researchers often worked in situations that were inherently familiar. Stenhouse also emphasized the need for evidence of rigorous processes and decisions in order to encourage robust practice and accountability to the wider field by allowing others to judge the quality of work through transparency of processes. Yin 1984 , the first book focused wholly on case study in research, gave a brief and basic outline of case study and associated practices. Various authors followed this approach, striving to engage more deeply in the significance of case study in the social sciences. Key among these are Merriam 1988 and Stake 1995 , along with Yin 1984 , who established powerful groundings for case study work. Additionally, evidence of the increasing popularity of case study can be found in a broad range of generic research methods texts, but these often do not have much scope for the extensive discussion of case study found in case study–specific books. Yin’s books and numerous editions provide a developing or evolving notion of case study with more detailed accounts of the possible purposes of case study, followed by Merriam 1988 and Stake 1995 who wrestled with alternative ways of looking at purposes and the positioning of case study within potential disciplinary modes. The authors referenced in this section are often characterized as the foundational authors on this subject and may have published various editions of their work, cited elsewhere in this article, based on their shifting ideas or emphases.

Merriam, S. B. 1988. Case study research in education: A qualitative approach . San Francisco: Jossey-Bass.

This is Merriam’s initial text on case study and is eminently accessible. The author establishes and reinforces various key features of case study; demonstrates support for positioning the case within a subject domain, e.g., psychology, sociology, etc.; and further shapes the case according to its purpose or intent.

Stake, R. E. 1995. The art of case study research . Thousand Oaks, CA: SAGE.

Stake is a very readable author, accessible and yet engaging with complex topics. The author establishes his key forms of case study: intrinsic, instrumental, and collective. Stake brings the reader through the process of conceptualizing the case, carrying it out, and analyzing the data. The author uses authentic examples to help readers understand and appreciate the nuances of an interpretive approach to case study.

Stenhouse, L. 1980. The study of samples and the study of cases. British Educational Research Journal 6:1–6.

DOI: 10.1080/0141192800060101

A key article in which Stenhouse sets out his stand on case study work. Those interested in the evolution of case study use in educational research should consider this article and the insights given.

Yin, R. K. 1984. Case Study Research: Design and Methods . Beverley Hills, CA: SAGE.

This preliminary text from Yin was very basic. However, it may be of interest in comparison with later books because Yin shows the ways in which case study as an approach or method in research has evolved in relation to detailed discussions of purpose, as well as the practicalities of working through the research process.

back to top

Users without a subscription are not able to see the full content on this page. Please subscribe or login .

Oxford Bibliographies Online is available by subscription and perpetual access to institutions. For more information or to contact an Oxford Sales Representative click here .

• About Education »
• Meet the Editorial Board »
• Academic Achievement
• Academic Audit for Universities
• Academic Freedom and Tenure in the United States
• Action Research in Education
• Adjuncts in Higher Education in the United States
• Administrator Preparation
• Adolescence
• Advanced Placement and International Baccalaureate Courses
• Advocacy and Activism in Early Childhood
• African American Racial Identity and Learning
• Alaska Native Education
• Alternative Certification Programs for Educators
• Alternative Schools
• American Indian Education
• Animals in Environmental Education
• Art Education
• Artificial Intelligence and Learning
• Assessing School Leader Effectiveness
• Assessment, Behavioral
• Assessment, Educational
• Assessment in Early Childhood Education
• Assistive Technology
• Augmented Reality in Education
• Beginning-Teacher Induction
• Bilingual Education and Bilingualism
• Black Undergraduate Women: Critical Race and Gender Perspe...
• Blended Learning
• Case Study in Education Research
• Changing Professional and Academic Identities
• Character Education
• Children’s and Young Adult Literature
• Children's Beliefs about Intelligence
• Children's Rights in Early Childhood Education
• Citizenship Education
• Civic and Social Engagement of Higher Education
• Classroom Learning Environments: Assessing and Investigati...
• Classroom Management
• Coherent Instructional Systems at the School and School Sy...
• College Admissions in the United States
• College Athletics in the United States
• Community Relations
• Comparative Education
• Computer-Assisted Language Learning
• Computer-Based Testing
• Conceptualizing, Measuring, and Evaluating Improvement Net...
• Continuous Improvement and "High Leverage" Educational Pro...
• Counseling in Schools
• Critical Approaches to Gender in Higher Education
• Critical Perspectives on Educational Innovation and Improv...
• Critical Race Theory
• Crossborder and Transnational Higher Education
• Cross-National Research on Continuous Improvement
• Cross-Sector Research on Continuous Learning and Improveme...
• Cultural Diversity in Early Childhood Education
• Culturally Responsive Leadership
• Culturally Responsive Pedagogies
• Culturally Responsive Teacher Education in the United Stat...
• Curriculum Design
• Data-driven Decision Making in the United States
• Deaf Education
• Desegregation and Integration
• Design Thinking and the Learning Sciences: Theoretical, Pr...
• Development, Moral
• Dialogic Pedagogy
• Digital Age Teacher, The
• Digital Citizenship
• Digital Divides
• Disabilities
• Distance Learning
• Distributed Leadership
• Doctoral Education and Training
• Early Childhood Education and Care (ECEC) in Denmark
• Early Childhood Education and Development in Mexico
• Early Childhood Education in Aotearoa New Zealand
• Early Childhood Education in Australia
• Early Childhood Education in China
• Early Childhood Education in Europe
• Early Childhood Education in Sub-Saharan Africa
• Early Childhood Education in Sweden
• Early Childhood Education Pedagogy
• Early Childhood Education Policy
• Early Childhood Education, The Arts in
• Early Childhood Mathematics
• Early Childhood Science
• Early Childhood Teacher Education
• Early Childhood Teachers in Aotearoa New Zealand
• Early Years Professionalism and Professionalization Polici...
• Economics of Education
• Education For Children with Autism
• Education for Sustainable Development
• Education Leadership, Empirical Perspectives in
• Education of Native Hawaiian Students
• Education Reform and School Change
• Educational Statistics for Longitudinal Research
• Educator Partnerships with Parents and Families with a Foc...
• Emotional and Affective Issues in Environmental and Sustai...
• Emotional and Behavioral Disorders
• Environmental and Science Education: Overlaps and Issues
• Environmental Education
• Environmental Education in Brazil
• Epistemic Beliefs
• Equity and Improvement: Engaging Communities in Educationa...
• Equity, Ethnicity, Diversity, and Excellence in Education
• Ethical Research with Young Children
• Ethics and Education
• Ethics of Teaching
• Ethnic Studies
• Evidence-Based Communication Assessment and Intervention
• Family and Community Partnerships in Education
• Family Day Care
• Federal Government Programs and Issues
• Feminization of Labor in Academia
• Finance, Education
• Financial Aid
• Formative Assessment
• Future-Focused Education
• Gender and Achievement
• Gender and Alternative Education
• Gender, Power and Politics in the Academy
• Gender-Based Violence on University Campuses
• Gifted Education
• Global Mindedness and Global Citizenship Education
• Global University Rankings
• Governance, Education
• Grounded Theory
• Growth of Effective Mental Health Services in Schools in t...
• Higher Education and Globalization
• Higher Education and the Developing World
• Higher Education Faculty Characteristics and Trends in the...
• Higher Education Finance
• Higher Education Governance
• Higher Education Graduate Outcomes and Destinations
• Higher Education in Africa
• Higher Education in China
• Higher Education in Latin America
• Higher Education in the United States, Historical Evolutio...
• Higher Education, International Issues in
• Higher Education Management
• Higher Education Policy
• Higher Education Research
• Higher Education Student Assessment
• High-stakes Testing
• History of Early Childhood Education in the United States
• History of Education in the United States
• History of Technology Integration in Education
• Homeschooling
• Inclusion in Early Childhood: Difference, Disability, and ...
• Inclusive Education
• Indigenous Education in a Global Context
• Indigenous Learning Environments
• Indigenous Students in Higher Education in the United Stat...
• Infant and Toddler Pedagogy
• Inservice Teacher Education
• Integrating Art across the Curriculum
• Intelligence
• Intensive Interventions for Children and Adolescents with ...
• International Perspectives on Academic Freedom
• Intersectionality and Education
• Knowledge Development in Early Childhood
• Leadership Development, Coaching and Feedback for
• Leadership in Early Childhood Education
• Leadership Training with an Emphasis on the United States
• Learning Analytics in Higher Education
• Learning Difficulties
• Learning, Lifelong
• Learning, Multimedia
• Learning Strategies
• Legal Matters and Education Law
• LGBT Youth in Schools
• Linguistic Diversity
• Linguistically Inclusive Pedagogy
• Literacy Development and Language Acquisition
• Literature Reviews
• Mathematics Identity
• Mathematics Instruction and Interventions for Students wit...
• Mathematics Teacher Education
• Measurement for Improvement in Education
• Measurement in Education in the United States
• Meta-Analysis and Research Synthesis in Education
• Methodological Approaches for Impact Evaluation in Educati...
• Methodologies for Conducting Education Research
• Mindfulness, Learning, and Education
• Motherscholars
• Multiliteracies in Early Childhood Education
• Multiple Documents Literacy: Theory, Research, and Applica...
• Multivariate Research Methodology
• Museums, Education, and Curriculum
• Music Education
• Narrative Research in Education
• Native American Studies
• Nonformal and Informal Environmental Education
• Note-Taking
• Numeracy Education
• One-to-One Technology in the K-12 Classroom
• Online Education
• Open Education
• Organizing for Continuous Improvement in Education
• Organizing Schools for the Inclusion of Students with Disa...
• Outdoor Play and Learning
• Outdoor Play and Learning in Early Childhood Education
• Pedagogical Leadership
• Pedagogy of Teacher Education, A
• Performance Objectives and Measurement
• Performance-based Research Assessment in Higher Education
• Performance-based Research Funding
• Phenomenology in Educational Research
• Philosophy of Education
• Physical Education
• Podcasts in Education
• Policy Context of United States Educational Innovation and...
• Politics of Education
• Portable Technology Use in Special Education Programs and ...
• Post-humanism and Environmental Education
• Pre-Service Teacher Education
• Problem Solving
• Productivity and Higher Education
• Professional Development
• Professional Learning Communities
• Programs and Services for Students with Emotional or Behav...
• Psychology Learning and Teaching
• Psychometric Issues in the Assessment of English Language ...
• Qualitative Data Analysis Techniques
• Qualitative, Quantitative, and Mixed Methods Research Samp...
• Qualitative Research Design
• Quantitative Research Designs in Educational Research
• Queering the English Language Arts (ELA) Writing Classroom
• Race and Affirmative Action in Higher Education
• Reading Education
• Refugee and New Immigrant Learners
• Relational and Developmental Trauma and Schools
• Relational Pedagogies in Early Childhood Education
• Reliability in Educational Assessments
• Religion in Elementary and Secondary Education in the Unit...
• Researcher Development and Skills Training within the Cont...
• Research-Practice Partnerships in Education within the Uni...
• Response to Intervention
• Restorative Practices
• Risky Play in Early Childhood Education
• Scale and Sustainability of Education Innovation and Impro...
• Scaling Up Research-based Educational Practices
• School Accreditation
• School Choice
• School Culture
• School District Budgeting and Financial Management in the ...
• School Improvement through Inclusive Education
• School Reform
• Schools, Private and Independent
• School-Wide Positive Behavior Support
• Science Education
• Secondary to Postsecondary Transition Issues
• Self-Regulated Learning
• Self-Study of Teacher Education Practices
• Service-Learning
• Severe Disabilities
• Single Salary Schedule
• Single-sex Education
• Single-Subject Research Design
• Social Context of Education
• Social Justice
• Social Network Analysis
• Social Pedagogy
• Social Science and Education Research
• Social Studies Education
• Sociology of Education
• Standards-Based Education
• Statistical Assumptions
• Student Access, Equity, and Diversity in Higher Education
• Student Assignment Policy
• Student Engagement in Tertiary Education
• Student Learning, Development, Engagement, and Motivation ...
• Student Participation
• Student Voice in Teacher Development
• Sustainability Education in Early Childhood Education
• Sustainability in Early Childhood Education
• Sustainability in Higher Education
• Teacher Beliefs and Epistemologies
• Teacher Collaboration in School Improvement
• Teacher Evaluation and Teacher Effectiveness
• Teacher Preparation
• Teacher Training and Development
• Teacher Unions and Associations
• Teacher-Student Relationships
• Teaching Critical Thinking
• Technologies, Teaching, and Learning in Higher Education
• Technology Education in Early Childhood
• Technology, Educational
• Technology-based Assessment
• The Bologna Process
• The Regulation of Standards in Higher Education
• Theories of Educational Leadership
• Three Conceptions of Literacy: Media, Narrative, and Gamin...
• Tracking and Detracking
• Traditions of Quality Improvement in Education
• Transformative Learning
• Transitions in Early Childhood Education
• Tribally Controlled Colleges and Universities in the Unite...
• Understanding the Psycho-Social Dimensions of Schools and ...
• University Faculty Roles and Responsibilities in the Unite...
• Using Ethnography in Educational Research
• Value of Higher Education for Students and Other Stakehold...
• Virtual Learning Environments
• Vocational and Technical Education
• Wellness and Well-Being in Education
• Women's and Gender Studies
• Young Children and Spirituality
• Young Children's Learning Dispositions
• Young Children's Working Theories
• Privacy Policy
• Cookie Policy
• Legal Notice
• Accessibility

Powered by:

• [66.249.64.20|81.177.182.174]
• 81.177.182.174

• Publications
• Conferences & Events
• Professional Learning
• Science Standards
• Awards & Competitions
• Daily Do Lesson Plans
• Free Resources
• American Rescue Plan
• For Preservice Teachers

NCCSTS Case Collection

• Partner Jobs in Education
• Interactive eBooks+
• Digital Catalog
• Regional Product Representatives
• e-Newsletters
• Bestselling Books
• Latest Books
• Popular Book Series
• Prospective Authors
• Web Seminars
• Exhibits & Sponsorship
• Conference Reviewers
• National Conference • Denver 24
• Leaders Institute 2024
• National Conference • New Orleans 24
• Submit a Proposal
• Latest Resources
• Professional Learning Units & Courses
• For Districts
• Online Course Providers
• Schools & Districts
• College Professors & Students
• The Standards
• Teachers and Admin
• eCYBERMISSION
• Toshiba/NSTA ExploraVision
• Junior Science & Humanities Symposium
• Teaching Awards
• Climate Change
• Earth & Space Science
• New Science Teachers
• Early Childhood
• Middle School
• High School
• Postsecondary
• Informal Education
• Journal Articles
• Lesson Plans
• e-newsletters
• Science & Children
• Science Scope
• The Science Teacher
• Journal of College Sci. Teaching
• Connected Science Learning
• NSTA Reports
• Next-Gen Navigator
• Science Update
• Teacher Tip Tuesday
• Trans. Sci. Learning

MyNSTA Community

• My Collections

Case Study Listserv

Permissions & Guidelines

Submit a Case Study

Resources & Publications

Enrich your students’ educational experience with case-based teaching

The NCCSTS Case Collection, created and curated by the National Center for Case Study Teaching in Science, on behalf of the University at Buffalo, contains over a thousand peer-reviewed case studies on a variety of topics in all areas of science.

Cases (only) are freely accessible; subscription is required for access to teaching notes and answer keys.

Subscribe Today

Browse Case Studies

Latest Case Studies

Development of the NCCSTS Case Collection was originally funded by major grants to the University at Buffalo from the National Science Foundation , The Pew Charitable Trusts , and the U.S. Department of Education .

• Case Studies

Teaching Guide

• Using the Open Case Studies Website
• Using the UBC Wiki
• Open Educational Resources
• Case Implementation
• Get Involved
• Process Documentation

Case studies offer a student-centered approach to learning that asks students to identify, explore, and provide solutions to real-world problems by focusing on case-specific examples (Wiek, Xiong, Brundiers, van der Leeuw, 2014, p 434). This approach simulates real life practice in sustainability education in that it illuminates the ongoing complexity of the problems being addressed. Publishing these case studies openly, means they can be re-used in a variety of contexts by others across campus and beyond. Since the cases never “end”; at any time students from all over UBC campus can engage with their content, highlighting their potential as powerful educational tools that can foster inter-disciplinary research of authentic problems. Students contributing to the case studies are making an authentic contribution to a deepening understanding of the complex challenges facing us in terms of environmental ethics and sustainability.

The case studies are housed on the UBC Wiki, and that content is then fed into the Open Case Studies website. The UBC Wiki as a platform for open, collaborative course work enables students to create, respond to and/or edit case studies, using the built in features (such as talk pages, document history and contributor track backs) to make editing transparent. The wiki also also helps students develop important transferable skills such as selection and curation of multimedia (while attending to copyright and re-use specifications), citation and referencing, summarizing research, etc. These activities help build critical thinking, creativity, collaboration, and digital literacy.

This guide is intended to help you get started with your case study project by offering:

• Information on how to use the UBC Wiki
• Research that supports case studies as effective tools for active learning
• Instructional strategies for teaching effectively with case studies
• Sample case study assignments used by UBC instructors

The UBC Wiki is a set of webpages accessible to anyone with a CWL account and has many unique features in addition to collaborative writing including the ability to revive previous drafts, and notifications setting that can support instructors in monitoring individual student contributions, or support students to better manage their collaborative efforts on their own. Using a wiki successfully in a course, however, requires proper facilitation and support from instructors and TAs.

The following links are helpful in getting started:

General Information:

• Navigating the Wiki: http://wiki.ubc.ca/Help:Navigation
• Wiki Help Table of Contents: http://wiki.ubc.ca/Help:Contents
• Frequently Asked Questions: http://wiki.ubc.ca/Help:Contents#Frequently_Asked_Questions

Self-Guided Wiki Tutorials:

• Getting Started With UBC Wiki - short video and links to common formatting needs.
• Beginner: http://wiki.ubc.ca/Documentation:MediaWiki_Basics/Learning_Activities/Beginner
• Intermediate: http://wiki.ubc.ca/Documentation:MediaWiki_Basics/Learning_Activities/Intermediate
• Advanced: http://wiki.ubc.ca/Documentation:MediaWiki_Basics/Learning_Activities/Advanced

The idea that learning is "active" is influenced by social constructivism , which emphasizes collaboration in the active co-construction of meaning among learners. Simply put, learning happens when people collaborate and interact with authentic learning tasks and situations. These ideas are becoming increasingly prevalent in the scholarly literature on teaching and learning (see for instance, Wilson 1996) and have important implications for pedagogy, especially in the university where traditional lectures remain the dominant instructional strategy. When students are asked to respond to authentic problems and questions, they assume responsibility for the trajectory of their learning, rather than it being decided upon by the instructor. This practice, also referred to as “student-centered learning” allows the students to become “active” participants in the construction of their understandings.

One of the easiest ways to develop higher order cognitive capacities (critical thinking, problem solving, creativity etc.) is through pedagogies that support inquiry based learning, thereby allowing students the opportunity to “develop [as] inquirers and to use curiosity, the urge to explore and understand...to become researchers and lifelong learners” (Justice, Rice, Roy, Hudspith & Jenkins,2009, p. 843). Because case studies are often collaborative, they provide unique inquiry based learning opportunities that will foster active engagement in student learning, while also teaching transferable skills (teamwork, collaboration, technology literacy). That the cases never “end” and that they can be considered by students and faculty from all over the UBC community, highlights their potential as powerful educational tools that can foster inter-disciplinary research of authentic problems.

Using case studies successfully in a course requires purposefully scaffolded support from the instructor and TA's. Instructors must properly introduce assignments, as well as facilitate and monitor the progress of students while they work on assignments. This will help ensure that students understand the purpose and value of the work they are doing and will also allow instructors and TA's to provide appropriate support and guidance.

The following instructional strategies will help you teach effectively using case studies:

1. Getting Started:

• Outline Your "Big Picture" Goals and Expectations : Communicate to students what you are hoping they will learn (Or have them tell you why they think you would ask them to work with case studies!). It is also important to discuss the quality of work you expect and offer specific examples of what that looks like. If you have any, look at exemplars of past student work, or simply evaluate existing case studies to generate a list of defining characteristics. Doing this will provide students with valuable tangible and visual examples of what you expect.
• Define "Case Study" : Don't assume that students understand what case-studies are, especially at the undergraduate level. Take the time to talk about what a case study is and why they are powerful teaching/learning tools. This can be facilitated during a tutorial with small group discussion. See Case Study Resources.
• Pick Case Studies Purposefully : If you are planning on having students evaluate case studies, make sure to read them in advance and have a clear understanding of why you chose it. This will help facilitate discussion and field student questions.
• Set the Context for the Evaluating or Creating the Case Study : Whether you are having students write the case studies themselves, or you are having them examine an existing case, it is important to set the parameters for how you want students to approach the problem. For instance, you may have them evaluate the case from the perspective of an industry professional, a community group or member, or even from their own perspective of university students. Whatever you choose, make sure you communicate this clearly.
• Set the Parameters for Evaluating or Creating the Case Study : Clearly outline all the information you want students to find out, and how you want it reported. You may want students to focus on some areas and disregard others, or you may want them to consider all the facts equally. Whatever you choose, make sure you communicate this clearly.

2. Use, Revise, and/or Create

• Use the case studies as they are : One way to use the case studies in courses is to have students read and discuss them as they are. They can be read on the open case studies website, downloaded from the wiki and embedded into another website, or downloaded in PDF or Microsoft Word format (see this guide for how to embed or download the case studies)
• If you are only making minor edits such as fixing a broken link or a typo, please go ahead. You could add a note about this to the "discussion" page to explain (see the tab at the top of each wiki page).
• You could add a section at the bottom of the case study with a perspective on it from your discipline. Some of the case studies already have sections at the bottom that are titled "What would a ___ do?" You can add a new one of those to give a different disciplinary perspective.
• If you want to make more substantial changes, it would be best if you copied and pasted the wiki content into a new page so as to preserve the original. The original version may be used in other courses by the instructor/students who created it, so making significant changes could be a problem! And those changes might be reverted by the original instructor and students (wiki pages keep all past versions, and those changes can easily be reverted). If you would like to substantially revise a case study, please contact Christina Hendricks, who can help you get started and then get the new version into the collection: [email protected]
• Create new case studies : We are always looking for new case studies for the collection! If you think you would like to write one, or involve your students in writing one, please contact Christina Hendricks: [email protected]

3. Guiding Case Study Discussions:

• Ask open-ended questions : Open-ended questions cannot be answered using "yes" or "no". Be careful when wording discussion questions, allowing them to be as open as possible.
• Listen Actively : Actively listen to students by paraphrasing what they have said to you and saying it back (e.g. "What I heard is....Is this what you meant?"). This will help you pay close attention to what they say and clarify any possible miscommunication.
• Role Play : Ask students to take on the perspective of different interested parties in considering the case study.
• Compare and Contrast : Ask students to compare and contrast cases in similar areas from the open case study collection. Discuss whether there are similar problems or possible solutions for the cases.

4. Staying on Track:

• Develop a Protocol for Collaboration : Have students outline how they will collaborate at the start of the assignment to ensure that the work is shared evenly and that each student has a purposeful role.
• Set Benchmark Assignments : Make sure students stay on track by requiring smaller assignments or assessments along the way. This can be as simple as coming to tutorial with a portion of the case-study written for peer critique and analysis.
• Give Students Adequate Time : Allow students enough time to read and consider case-studies thoughtfully. The more time you can provide, the less overwhelmed students will feel. This will encourage them to go deeper with their case study and their learning.
• Forestry : In this assignment, students in a graduate course wrote their own case studies. This link provides information on the assignment, a handout given to the students, and a grading rubric: Short-Term Assignment: What is Illegal Logging? - Teacher Guide
• Political Science : Students in a third-year political science class responded to a case study written by the instructor. They worked in groups to create action plans for climate change problems. This link provides information on the assignment as well as a handout given to the students: Class Activity: Action Plans for Climate Change - Teacher Guide
• Education : Teacher candidates in the Faculty of Education respond to case studies written by students. They discuss a case study and respond to questions with the goal of identifying the issues raised, perspectives involved and possible ways forward. The goal is to support decision making related to online presence and social media engagement. Digital Tattoo Case Studies for Student Teachers Facilitators' Guide

• High contrast
• Press Centre

Search UNICEF

Education case studies, around-the-world case studies on unicef's education programme.

Education knowledge management dashboard

Case studies

Adolescent education and skills.

Improving students' mental health in Bangladesh

Improving the quality of lower secondary through inquiry-based learning and skills development (Argentina)

An online career portal strengthens career guidance among secondary students in India and helps them plan for future educational and work opportunities (India)

Lessons on youth-led action towards climate advocacy and policy (India)

Learning, life skills and citizenship education and social cohesion through game-based sports – Nashatati Programme (Jordan)

Mental health promotion and suicide prevention in schools (Kazakhstan)

A multi-level, cross-sectoral response to improving adolescent mental health (Mongolia)

The Personal Project (Morocco)  

Improving adolescents’ learning in violence-affected areas through blended in-person and online learning opportunities - Communities in Harmony for Children and Adolescents (Mexico)

A community-based approach to support the psychosocial wellbeing of students and teachers (Nicaragua)

Flexible pathways help build the skills and competencies of vulnerable out-of-school adolescents (United Republic of Tanzania)

Climate change and education

Schools as platforms for climate action (Cambodia)

Paving the way for a climate resilient education system (India)

Youth act against climate and air pollution impacts (Mongolia)

Early childhood education

Early environments of care: Strengthening the foundation of children’s development, mental health and wellbeing (Bhutan)

Native language education paves the way for preschool readiness (Bolivia)

Developing cross-sector quality standards for children aged 0-7 (Bulgaria)

Expanding quality early learning through results-based financing (Cambodia)

Harnessing technology to promote communication, education and social inclusion for young children with developmental delays and disabilities (Croatia, Montenegro, and Serbia)

Scaling up quality early childhood education in India by investing in ongoing professional development for officials at the state, district and local levels (India)

Strengthening early childhood education in the national education plan and budget in Lesotho to help children succeed in primary and beyond (Lesotho)

Enhancing play-based learning through supportive supervision (Nigeria)

Learning social and emotional skills in pre-school creates brighter futures for children (North Macedonia)

How developing minimum standards increased access to pre-primary education (Rwanda)

Expanding access to quality early childhood education for the most excluded children (Serbia)

Advancing early learning through results-based financing (Sierra Leone)

Lessons learned from designing social impact bonds to expand preschool education (Uzbekistan)

Equity and inclusion

Inclusive education for children with disabilities.

Strengthening policies to mainstream disability inclusion in pre-primary education (Ethiopia)

National early screening and referrals are supporting more young children with disabilities to learn (Jamaica)

Ensuring inclusive education during the pandemic and beyond (Dominican Republic)

Championing inclusive practices for children with disabilities (Ghana)

Accessible digital textbooks for children in Kenya (Kenya)

Planning for inclusion (Nepal)

Harnessing the potential of inclusive digital education to improve learning (Paraguay)

Gender equality in education

Sparking adolescent girls' participation and interest in STEM (Ghana)

Non-formal education and the use of data and evidence help marginalized girls learn in Nepal (Nepal)

Getting girls back to the classroom after COVID-19 school closures (South Sudan)

Education in emergencies

Creating classrooms that are responsive to the mental health needs of learners, including refugees (Poland)

Return to school (Argentina)

Learning from the education sector’s COVID-19 response to prepare for future emergencies (Bangladesh)

Prioritising learning for Rohingya children (Bangladesh)

Prioritizing children and adolescents’ mental health and protection during school reopening (Brazil)

Learning where it is difficult to learn: Radio programmes help keep children learning in Cameroon

Reaching the final mile for all migrant children to access education (Colombia)

Supporting the learning and socio-emotional development of refugee children (Colombia)

Mission Recovery (Democratic Republic of the Congo)

The National Building the Foundations for Learning Program, CON BASE (Dominican Republic)

Mental health and psychosocial well-being services are integrated in the education system (Ecuador)

Improving access to quality education for refugee learners (Ethiopia)

The Learning Passport and non-formal education for vulnerable children and youth (Lebanon)

Accelerated Learning Programme improves children’s learning in humanitarian settings (Mozambique)

Responding to multiple emergencies – building teachers’ capacity to provide mental health and psychosocial support before, during, and after crises (Mozambique)

Teaching at the right level to improve learning in Borno State (Nigeria)

Remedial catch-up learning programmes support children with COVID-19 learning loss and inform the national foundational learning strategy (Rwanda)

Learning solutions for pastoralist and internally displaced children (Somalia)

Recovering learning at all levels (South Africa)

How radio education helped children learn during the COVID-19 pandemic and aftermath (South Sudan)

Addressing learning loss through EiE and remedial education for children in Gaza (State of Palestine)

Providing psychosocial support and promoting learning readiness during compounding crises for adolescents in Gaza (State of Palestine)

Inclusion of South Sudanese refugees into the national education system (Sudan)

Inclusion of Syrian refugee children into the national education system (Turkey)

Including refugee learners so that every child learns (Uganda)

Learning assessments

Assessment for learning (Afghanistan)

Formative assessment places student learning at the heart of teaching (Ethiopia)

Strengthening teacher capacity for formative assessment (Europe and Central Asia)

All students back to learning (India)

Strengthening the national assessment system through the new National Achievement Survey improves assessment of children’s learning outcomes (India)

A new phone-based learning assessment targets young children (Nepal)

Adapting a remote platform in innovative ways to assess learning (Nigeria)

Assessing children's reading in indigenous languages (Peru)

Southeast Asia primary learning metrics: Assessing the learning outcomes of grade 5 students (Southeast Asia)

Minimising learning gaps among early-grade learners (Sri Lanka)

Assessing early learning (West and Central Africa)

Primary education / Foundational Literacy and Numeracy

Supporting Teachers to Improve Foundational Learning for Syrian Refugee Students in Jordan

Empowering teachers in Guinea: Transformative solutions for foundational learning

Improving child and adolescent health and nutrition through policy advocacy (Argentina)

Online diagnostic testing and interactive tutoring (Bulgaria)

Supporting the socio-emotional learning and psychological wellbeing of children through a whole-school approach (China)

Engaging parents to overcome reading poverty (India)

Integrated school health and wellness ensure better learning for students (India)

Instruction tailored to students’ learning levels improves literacy (Indonesia)

A whole-school approach to improve learning, safety and wellbeing (Jamaica)

Multi-sectoral programme to improve the nutrition of school-aged adolescents (Malawi)

Parents on the frontlines of early grade reading and math (Nigeria)

Training, inspiring and motivating early grade teachers to strengthen children’s skills in literacy and numeracy (Sierra Leone) Life skills and citizenship education through Experiential Learning Objects Bank (State of Palestine)

Curriculum reform to meet the individual needs of students (Uzbekistan)

Improving early grade reading and numeracy through ‘Catch-Up,’ a remedial learning programme (Zambia)

Reimagine Education / Digital learning

Education 2.0: skills-based education and digital learning (Egypt)

Empowering adolescents through co-creation of innovative digital solutions (Indonesia)

Virtual instructional leadership course (Jamaica)

Learning Bridges accelerates learning for over 600,000 students (Jordan)

Unleashing the potential of youth through the Youth Learning Passport (Jordan)

Lessons learned from the launch of the Learning Passport Shkollat.org (Kosovo)

Opening up the frontiers of digital learning with the Learning Passport (Lao PDR)

Building teachers’ confidence and capacity to provide online learning (Maldives)

Mauritania’s first digital learning program: Akelius Digital French Course (Mauritania)

Mitigating learning loss and strengthening foundational skills through the Learning Passport (Mexico)

Expanding digital learning opportunities and connectivity for all learners (Tajikistan)

For COVID-19 education case studies, please click here and filter by area of work (Education) and type (Case Study / Field Notes).

Resources for partners

Learning at the heart of education

Key Asks 2021 - National Reviews - SDG 4 Quality Education

More from UNICEF

Transforming education in africa.

An evidence-based overview and recommendations for long-term improvements

Early Childhood Education for All

It is time for a world where all children enter school equipped with the skills they need to succeed.

A world ready to learn

Prioritizing quality early childhood education

Mission: Recovery education in humanitarian countries

Updates on UNICEF’s work to deliver education to children in crisis-affected countries, with support from the US Government

• Board of Directors
• National Advisory Board
• OUR RESOURCES
• OUR STORIES
• Our Newsletters

Case Study Compilation

The SEL Integration Approach  Case Study Compilation  was developed with and for educators who work in a K-12 school setting, including teachers, paraprofessionals, counselors, SEL Directors, teacher leaders, & school principals, to provide examples of practice related to three questions:

• What does it mean to focus on social-emotional development and the creation of positive learning environments?
• How can educators integrate their approaches to social, emotional, and academic development?
• What does it look, sound, and feel like when SEL is effectively embedded into all elements of the school day?

When read one at a time, the case studies offer snapshots of social-emotional learning in action; they describe daily routines, activities, and teachable moments within short vignettes. When read together, the case studies provide a unique picture of what it takes for a school to integrate social, emotional, and academic learning across grade levels, content areas, and other unique contexts.

The Case Study Compilation includes:

• Eleven case studies:  Each case study highlights educator ‘moves’ and strategies to embed social-emotional skills, mindsets, and competencies throughout the school day and within academics. They each  conclude with a reflection prompt that challenges readers to examine their own practice. The case studies are written from several different perspectives, including teachers in the classroom and in distance learning environments, a school counselor, and district leaders.
• Reflection Guide for Professional Learning:  The Reflection Guide offers an entry point for educators to think critically about their work with youth in order to strengthen their practice. School leaders or other partners may choose to use this Reflection Guide in a variety of contexts, including coaching conversations and staff professional development sessions.

View our accompanying Quick Reference Guide , Companion Guides , and Educator & School Leader Self-Reflection Tools .

“We must resist thinking in siloed terms when it comes to social-emotional learning (SEL), academics, and equity. Rather, these elements of our work as educators and partners go hand in hand.”

HEAD & HEART, TransformEd & ANet

Get the Latest Updates

© Copyright 2020 Transforming Education. All content is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License .

Transforming Education is a registered 501(c)3 based in Boston, Massachusetts

Recent Tweet

Take action.

22 Cases and Articles to Help Bring Diversity Issues into Class Discussions

Explore more.

• Course Materials
• Diversity, Equity, and Inclusion

T he recent civic unrest in the United States following the death of George Floyd has elevated the urgency to recognize and study issues of diversity and the needs of underrepresented groups in all aspects of public life.

Business schools—and educational institutions across the spectrum—are no exception. It’s vital that educators facilitate safe and productive dialogue with students about issues of inclusion and diversity. To help, we’ve gathered a collection of case studies (all with teaching notes) and articles that can encourage and support these critical discussions.

These materials are listed across three broad topic areas: leadership and inclusion, cases featuring protagonists from historically underrepresented groups, and women and leadership around the world. This list is hardly exhaustive, but we hope it provides ways to think creatively and constructively about how educators can integrate these important topics in their classes. HBP will continue to curate and share content that addresses these equity issues and that features diverse protagonists.

Editors’ note: To access the full text of these articles, cases, and accompanying teaching notes, you must be registered with HBP Education. We invite you to sign up for a free educator account here . Verification may take a day; in the meantime, you can read all of our Inspiring Minds content .

Leadership and Inclusion

John Rogers, Jr.—Ariel Investments Co.

—by Steven S. Rogers and Greg White

Gender and Free Speech at Google (A)

—by Nien-hê Hsieh, Martha J. Crawford, and Sarah Mehta

The Massport Model: Integrating Diversity and Inclusion into Public-Private Partnerships

—by Laura Winig and Robert Livingston

“Numbers Take Us Only So Far”

—by Maxine Williams

For Women and Minorities to Get Ahead, Managers Must Assign Work Fairly

—by Joan C. Williams and Marina Multhaup

How Organizations Are Failing Black Workers—and How to Do Better

—by Adia Harvey Wingfield

To Retain Employees, Focus on Inclusion—Not Just Diversity

—by Karen Brown

From HBR 's The Big Idea:

Toward a Racially Just Workplace: Diversity efforts are failing black employees. Here’s a better approach.

—by Laura Morgan Roberts and Anthony J. Mayo

Cases with Protagonists from Historically Underrepresented Groups

Arlan Hamilton and Backstage Capital

—by Laura Huang and Sarah Mehta

United Housing—Otis Gates

—by Steven Rogers and Mercer Cook

Eve Hall: The African American Investment Fund in Milwaukee

—by Steven Rogers and Alterrell Mills

Dylan Pierce at Peninsula Industries

—by Karthik Ramanna

Maggie Lena Walker and the Independent Order of St. Luke

—by Anthony J. Mayo and Shandi O. Smith

Multimedia Cases:

Enterprise Risk Management at Hydro One, Multimedia Case

—by Anette Mikes

Women and Leadership Around the World

Monique Leroux: Leading Change at Desjardins

—by Rosabeth Moss Kanter and Ai-Ling Jamila Malone

Kaweyan: Female Entrepreneurship and the Past and Future of Afghanistan

—by Geoffrey G. Jones and Gayle Tzemach Lemmon

Womenomics in Japan

—by Boris Groysberg, Mayuka Yamazaki, Nobuo Sato, and David Lane

Women MBAs at Harvard Business School: 1962-2012

—by Boris Groysberg, Kerry Herman, and Annelena Lobb

Beating the Odds

—by Laura Morgan Roberts, Anthony J. Mayo, Robin J. Ely, and David A. Thomas

Rethink What You “Know” About High-Achieving Women

—by Robin J. Ely, Pamela Stone, and Colleen Ammerman

“I Try to Spark New Ideas”

—by Christine Lagarde and Adi Ignatius

How Women Manage the Gendered Norms of Leadership

—by Wei Zheng, Ronit Kark, and Alyson Meister

Is this list helpful to you? What other topics or materials would you like to see featured in our next curated list? Let us know .

Related Articles

We use cookies to understand how you use our site and to improve your experience, including personalizing content. Learn More . By continuing to use our site, you accept our use of cookies and revised Privacy Policy .

Writing A Case Study

Case Study Examples

Brilliant Case Study Examples and Templates For Your Help

15 min read

People also read

A Complete Case Study Writing Guide With Examples

Simple Case Study Format for Students to Follow

Understand the Types of Case Study Here

It’s no surprise that writing a case study is one of the most challenging academic tasks for students. You’re definitely not alone here!

Most people don't realize that there are specific guidelines to follow when writing a case study. If you don't know where to start, it's easy to get overwhelmed and give up before you even begin.

Don't worry! Let us help you out!

We've collected over 25 free case study examples with solutions just for you. These samples with solutions will help you win over your panel and score high marks on your case studies.

So, what are you waiting for? Let's dive in and learn the secrets to writing a successful case study.

• 1. An Overview of Case Studies
• 2. Case Study Examples for Students
• 3. Business Case Study Examples
• 4. Medical Case Study Examples
• 5. Psychology Case Study Examples 
• 6. Sales Case Study Examples
• 7. Interview Case Study Examples
• 8. Marketing Case Study Examples
• 9. Tips to Write a Good Case Study

An Overview of Case Studies

A case study is a research method used to study a particular individual, group, or situation in depth. It involves analyzing and interpreting data from a variety of sources to gain insight into the subject being studied. 

Case studies are often used in psychology, business, and education to explore complicated problems and find solutions. They usually have detailed descriptions of the subject, background info, and an analysis of the main issues.

The goal of a case study is to provide a comprehensive understanding of the subject. Typically, case studies can be divided into three parts, challenges, solutions, and results. 

Here is a case study sample PDF so you can have a clearer understanding of what a case study actually is:

Case Study Sample PDF

How to Write a Case Study Examples

Learn how to write a case study with the help of our comprehensive case study guide.

Case Study Examples for Students

Quite often, students are asked to present case studies in their academic journeys. The reason instructors assign case studies is for students to sharpen their critical analysis skills, understand how companies make profits, etc.

Below are some case study examples in research, suitable for students:

Case Study Example in Software Engineering

Qualitative Research Case Study Sample

Software Quality Assurance Case Study

Social Work Case Study Example

Ethical Case Study

Case Study Example PDF

These examples can guide you on how to structure and format your own case studies.

Struggling with formatting your case study? Check this case study format guide and perfect your document’s structure today.

Business Case Study Examples

A business case study examines a business’s specific challenge or goal and how it should be solved. Business case studies usually focus on several details related to the initial challenge and proposed solution. 

To help you out, here are some samples so you can create case studies that are related to businesses: 

Here are some more business case study examples:

Business Case Studies PDF

Business Case Studies Example

Typically, a business case study discovers one of your customer's stories and how you solved a problem for them. It allows your prospects to see how your solutions address their needs. 

Medical Case Study Examples

Medical case studies are an essential part of medical education. They help students to understand how to diagnose and treat patients. 

Here are some medical case study examples to help you.

Medical Case Study Example

Nursing Case Study Example

Want to understand the various types of case studies? Check out our types of case study blog to select the perfect type.

Psychology Case Study Examples 

Case studies are a great way of investigating individuals with psychological abnormalities. This is why it is a very common assignment in psychology courses. 

By examining all the aspects of your subject’s life, you discover the possible causes of exhibiting such behavior. 

For your help, here are some interesting psychology case study examples:

Psychology Case Study Example

Mental Health Case Study Example

Sales Case Study Examples

Case studies are important tools for sales teams’ performance improvement. By examining sales successes, teams can gain insights into effective strategies and create action plans to employ similar tactics.

By researching case studies of successful sales campaigns, sales teams can more accurately identify challenges and develop solutions.

Sales Case Study Example

Interview Case Study Examples

Interview case studies provide businesses with invaluable information. This data allows them to make informed decisions related to certain markets or subjects.

Interview Case Study Example

Marketing Case Study Examples

Marketing case studies are real-life stories that showcase how a business solves a problem. They typically discuss how a business achieves a goal using a specific marketing strategy or tactic.

They typically describe a challenge faced by a business, the solution implemented, and the results achieved.

This is a short sample marketing case study for you to get an idea of what an actual marketing case study looks like.

 Here are some more popular marketing studies that show how companies use case studies as a means of marketing and promotion:

“Chevrolet Discover the Unexpected” by Carol H. Williams

This case study explores Chevrolet's “ DTU Journalism Fellows ” program. The case study uses the initials “DTU” to generate interest and encourage readers to learn more. 

Multiple types of media, such as images and videos, are used to explain the challenges faced. The case study concludes with an overview of the achievements that were met.

Key points from the case study include:

• Using a well-known brand name in the title can create interest.
• Combining different media types, such as headings, images, and videos, can help engage readers and make the content more memorable.
• Providing a summary of the key achievements at the end of the case study can help readers better understand the project's impact.

“The Met” by Fantasy

“ The Met ” by Fantasy is a fictional redesign of the Metropolitan Museum of Art in New York City, created by the design studio Fantasy. The case study clearly and simply showcases the museum's website redesign.

The Met emphasizes the website’s features and interface by showcasing each section of the interface individually, allowing the readers to concentrate on the significant elements.

For those who prefer text, each feature includes an objective description. The case study also includes a “Contact Us” call-to-action at the bottom of the page, inviting visitors to contact the company.

Key points from this “The Met” include:

• Keeping the case study simple and clean can help readers focus on the most important aspects.
• Presenting the features and solutions with a visual showcase can be more effective than writing a lot of text.
• Including a clear call-to-action at the end of the case study can encourage visitors to contact the company for more information.

“Better Experiences for All” by Herman Miller

Herman Miller's minimalist approach to furniture design translates to their case study, “ Better Experiences for All ”, for a Dubai hospital. The page features a captivating video with closed-captioning and expandable text for accessibility.

The case study presents a wealth of information in a concise format, enabling users to grasp the complexities of the strategy with ease. It concludes with a client testimonial and a list of furniture items purchased from the brand.

Key points from the “Better Experiences” include:

• Make sure your case study is user-friendly by including accessibility features like closed captioning and expandable text.
• Include a list of products that were used in the project to guide potential customers.

“NetApp” by Evisort 

Evisort's case study on “ NetApp ” stands out for its informative and compelling approach. The study begins with a client-centric overview of NetApp, strategically directing attention to the client rather than the company or team involved.

The case study incorporates client quotes and explores NetApp’s challenges during COVID-19. Evisort showcases its value as a client partner by showing how its services supported NetApp through difficult times. 

• Provide an overview of the company in the client’s words, and put focus on the customer. 
• Highlight how your services can help clients during challenging times.
• Make your case study accessible by providing it in various formats.

“Red Sox Season Campaign,” by CTP Boston

The “ Red Sox Season Campaign ” showcases a perfect blend of different media, such as video, text, and images. Upon visiting the page, the video plays automatically, there are videos of Red Sox players, their images, and print ads that can be enlarged with a click.

The page features an intuitive design and invites viewers to appreciate CTP's well-rounded campaign for Boston's beloved baseball team. There’s also a CTA that prompts viewers to learn how CTP can create a similar campaign for their brand.

Some key points to take away from the “Red Sox Season Campaign”: 

• Including a variety of media such as video, images, and text can make your case study more engaging and compelling.
• Include a call-to-action at the end of your study that encourages viewers to take the next step towards becoming a customer or prospect.

“Airbnb + Zendesk” by Zendesk

The case study by Zendesk, titled “ Airbnb + Zendesk : Building a powerful solution together,” showcases a true partnership between Airbnb and Zendesk. 

The article begins with an intriguing opening statement, “Halfway around the globe is a place to stay with your name on it. At least for a weekend,” and uses stunning images of beautiful Airbnb locations to captivate readers.

Instead of solely highlighting Zendesk's product, the case study is crafted to tell a good story and highlight Airbnb's service in detail. This strategy makes the case study more authentic and relatable.

Some key points to take away from this case study are:

• Use client's offerings' images rather than just screenshots of your own product or service.
• To begin the case study, it is recommended to include a distinct CTA. For instance, Zendesk presents two alternatives, namely to initiate a trial or seek a solution.

“Influencer Marketing” by Trend and WarbyParker

The case study "Influencer Marketing" by Trend and Warby Parker highlights the potential of influencer content marketing, even when working with a limited budget. 

The “Wearing Warby” campaign involved influencers wearing Warby Parker glasses during their daily activities, providing a glimpse of the brand's products in use. 

This strategy enhanced the brand's relatability with influencers' followers. While not detailing specific tactics, the case study effectively illustrates the impact of third-person case studies in showcasing campaign results.

Key points to take away from this case study are:

• Influencer marketing can be effective even with a limited budget.
• Showcasing products being used in everyday life can make a brand more approachable and relatable.
• Third-person case studies can be useful in highlighting the success of a campaign.

Marketing Case Study Example

Marketing Case Study Template

Now that you have read multiple case study examples, hop on to our tips.

Tips to Write a Good Case Study

Here are some note-worthy tips to craft a winning case study 

• Define the purpose of the case study This will help you to focus on the most important aspects of the case. The case study objective helps to ensure that your finished product is concise and to the point.
• Choose a real-life example. One of the best ways to write a successful case study is to choose a real-life example. This will give your readers a chance to see how the concepts apply in a real-world setting.
• Keep it brief. This means that you should only include information that is directly relevant to your topic and avoid adding unnecessary details.
• Use strong evidence. To make your case study convincing, you will need to use strong evidence. This can include statistics, data from research studies, or quotes from experts in the field.
• Edit and proofread your work. Before you submit your case study, be sure to edit and proofread your work carefully. This will help to ensure that there are no errors and that your paper is clear and concise.

There you go!

We’re sure that now you have secrets to writing a great case study at your fingertips! This blog teaches the key guidelines of various case studies with samples. So grab your pen and start crafting a winning case study right away!

Having said that, we do understand that some of you might be having a hard time writing compelling case studies.

But worry not! Our expert case study writing service is here to take all your case-writing blues away! 

With 100% thorough research guaranteed, our online essay service can craft an amazing case study within 24 hours! 

So why delay? Let us help you shine in the eyes of your instructor!

Write Essay Within 60 Seconds!

Dr. Barbara is a highly experienced writer and author who holds a Ph.D. degree in public health from an Ivy League school. She has worked in the medical field for many years, conducting extensive research on various health topics. Her writing has been featured in several top-tier publications.

Paper Due? Why Suffer? That’s our Job!

Keep reading

• Privacy Policy

Home » Case Study – Methods, Examples and Guide

Case Study – Methods, Examples and Guide

Table of Contents

A case study is a research method that involves an in-depth examination and analysis of a particular phenomenon or case, such as an individual, organization, community, event, or situation.

It is a qualitative research approach that aims to provide a detailed and comprehensive understanding of the case being studied. Case studies typically involve multiple sources of data, including interviews, observations, documents, and artifacts, which are analyzed using various techniques, such as content analysis, thematic analysis, and grounded theory. The findings of a case study are often used to develop theories, inform policy or practice, or generate new research questions.

Types of Case Study

Types and Methods of Case Study are as follows:

Single-Case Study

A single-case study is an in-depth analysis of a single case. This type of case study is useful when the researcher wants to understand a specific phenomenon in detail.

For Example , A researcher might conduct a single-case study on a particular individual to understand their experiences with a particular health condition or a specific organization to explore their management practices. The researcher collects data from multiple sources, such as interviews, observations, and documents, and uses various techniques to analyze the data, such as content analysis or thematic analysis. The findings of a single-case study are often used to generate new research questions, develop theories, or inform policy or practice.

Multiple-Case Study

A multiple-case study involves the analysis of several cases that are similar in nature. This type of case study is useful when the researcher wants to identify similarities and differences between the cases.

For Example, a researcher might conduct a multiple-case study on several companies to explore the factors that contribute to their success or failure. The researcher collects data from each case, compares and contrasts the findings, and uses various techniques to analyze the data, such as comparative analysis or pattern-matching. The findings of a multiple-case study can be used to develop theories, inform policy or practice, or generate new research questions.

Exploratory Case Study

An exploratory case study is used to explore a new or understudied phenomenon. This type of case study is useful when the researcher wants to generate hypotheses or theories about the phenomenon.

For Example, a researcher might conduct an exploratory case study on a new technology to understand its potential impact on society. The researcher collects data from multiple sources, such as interviews, observations, and documents, and uses various techniques to analyze the data, such as grounded theory or content analysis. The findings of an exploratory case study can be used to generate new research questions, develop theories, or inform policy or practice.

Descriptive Case Study

A descriptive case study is used to describe a particular phenomenon in detail. This type of case study is useful when the researcher wants to provide a comprehensive account of the phenomenon.

For Example, a researcher might conduct a descriptive case study on a particular community to understand its social and economic characteristics. The researcher collects data from multiple sources, such as interviews, observations, and documents, and uses various techniques to analyze the data, such as content analysis or thematic analysis. The findings of a descriptive case study can be used to inform policy or practice or generate new research questions.

Instrumental Case Study

An instrumental case study is used to understand a particular phenomenon that is instrumental in achieving a particular goal. This type of case study is useful when the researcher wants to understand the role of the phenomenon in achieving the goal.

For Example, a researcher might conduct an instrumental case study on a particular policy to understand its impact on achieving a particular goal, such as reducing poverty. The researcher collects data from multiple sources, such as interviews, observations, and documents, and uses various techniques to analyze the data, such as content analysis or thematic analysis. The findings of an instrumental case study can be used to inform policy or practice or generate new research questions.

Case Study Data Collection Methods

Here are some common data collection methods for case studies:

Interviews involve asking questions to individuals who have knowledge or experience relevant to the case study. Interviews can be structured (where the same questions are asked to all participants) or unstructured (where the interviewer follows up on the responses with further questions). Interviews can be conducted in person, over the phone, or through video conferencing.

Observations

Observations involve watching and recording the behavior and activities of individuals or groups relevant to the case study. Observations can be participant (where the researcher actively participates in the activities) or non-participant (where the researcher observes from a distance). Observations can be recorded using notes, audio or video recordings, or photographs.

Documents can be used as a source of information for case studies. Documents can include reports, memos, emails, letters, and other written materials related to the case study. Documents can be collected from the case study participants or from public sources.

Surveys involve asking a set of questions to a sample of individuals relevant to the case study. Surveys can be administered in person, over the phone, through mail or email, or online. Surveys can be used to gather information on attitudes, opinions, or behaviors related to the case study.

Artifacts are physical objects relevant to the case study. Artifacts can include tools, equipment, products, or other objects that provide insights into the case study phenomenon.

How to conduct Case Study Research

Conducting a case study research involves several steps that need to be followed to ensure the quality and rigor of the study. Here are the steps to conduct case study research:

• Define the research questions: The first step in conducting a case study research is to define the research questions. The research questions should be specific, measurable, and relevant to the case study phenomenon under investigation.
• Select the case: The next step is to select the case or cases to be studied. The case should be relevant to the research questions and should provide rich and diverse data that can be used to answer the research questions.
• Collect data: Data can be collected using various methods, such as interviews, observations, documents, surveys, and artifacts. The data collection method should be selected based on the research questions and the nature of the case study phenomenon.
• Analyze the data: The data collected from the case study should be analyzed using various techniques, such as content analysis, thematic analysis, or grounded theory. The analysis should be guided by the research questions and should aim to provide insights and conclusions relevant to the research questions.
• Draw conclusions: The conclusions drawn from the case study should be based on the data analysis and should be relevant to the research questions. The conclusions should be supported by evidence and should be clearly stated.
• Validate the findings: The findings of the case study should be validated by reviewing the data and the analysis with participants or other experts in the field. This helps to ensure the validity and reliability of the findings.
• Write the report: The final step is to write the report of the case study research. The report should provide a clear description of the case study phenomenon, the research questions, the data collection methods, the data analysis, the findings, and the conclusions. The report should be written in a clear and concise manner and should follow the guidelines for academic writing.

Examples of Case Study

Here are some examples of case study research:

• The Hawthorne Studies : Conducted between 1924 and 1932, the Hawthorne Studies were a series of case studies conducted by Elton Mayo and his colleagues to examine the impact of work environment on employee productivity. The studies were conducted at the Hawthorne Works plant of the Western Electric Company in Chicago and included interviews, observations, and experiments.
• The Stanford Prison Experiment: Conducted in 1971, the Stanford Prison Experiment was a case study conducted by Philip Zimbardo to examine the psychological effects of power and authority. The study involved simulating a prison environment and assigning participants to the role of guards or prisoners. The study was controversial due to the ethical issues it raised.
• The Challenger Disaster: The Challenger Disaster was a case study conducted to examine the causes of the Space Shuttle Challenger explosion in 1986. The study included interviews, observations, and analysis of data to identify the technical, organizational, and cultural factors that contributed to the disaster.
• The Enron Scandal: The Enron Scandal was a case study conducted to examine the causes of the Enron Corporation’s bankruptcy in 2001. The study included interviews, analysis of financial data, and review of documents to identify the accounting practices, corporate culture, and ethical issues that led to the company’s downfall.
• The Fukushima Nuclear Disaster : The Fukushima Nuclear Disaster was a case study conducted to examine the causes of the nuclear accident that occurred at the Fukushima Daiichi Nuclear Power Plant in Japan in 2011. The study included interviews, analysis of data, and review of documents to identify the technical, organizational, and cultural factors that contributed to the disaster.

Application of Case Study

Case studies have a wide range of applications across various fields and industries. Here are some examples:

Business and Management

Case studies are widely used in business and management to examine real-life situations and develop problem-solving skills. Case studies can help students and professionals to develop a deep understanding of business concepts, theories, and best practices.

Case studies are used in healthcare to examine patient care, treatment options, and outcomes. Case studies can help healthcare professionals to develop critical thinking skills, diagnose complex medical conditions, and develop effective treatment plans.

Case studies are used in education to examine teaching and learning practices. Case studies can help educators to develop effective teaching strategies, evaluate student progress, and identify areas for improvement.

Social Sciences

Case studies are widely used in social sciences to examine human behavior, social phenomena, and cultural practices. Case studies can help researchers to develop theories, test hypotheses, and gain insights into complex social issues.

Law and Ethics

Case studies are used in law and ethics to examine legal and ethical dilemmas. Case studies can help lawyers, policymakers, and ethical professionals to develop critical thinking skills, analyze complex cases, and make informed decisions.

Purpose of Case Study

The purpose of a case study is to provide a detailed analysis of a specific phenomenon, issue, or problem in its real-life context. A case study is a qualitative research method that involves the in-depth exploration and analysis of a particular case, which can be an individual, group, organization, event, or community.

The primary purpose of a case study is to generate a comprehensive and nuanced understanding of the case, including its history, context, and dynamics. Case studies can help researchers to identify and examine the underlying factors, processes, and mechanisms that contribute to the case and its outcomes. This can help to develop a more accurate and detailed understanding of the case, which can inform future research, practice, or policy.

Case studies can also serve other purposes, including:

• Illustrating a theory or concept: Case studies can be used to illustrate and explain theoretical concepts and frameworks, providing concrete examples of how they can be applied in real-life situations.
• Developing hypotheses: Case studies can help to generate hypotheses about the causal relationships between different factors and outcomes, which can be tested through further research.
• Providing insight into complex issues: Case studies can provide insights into complex and multifaceted issues, which may be difficult to understand through other research methods.
• Informing practice or policy: Case studies can be used to inform practice or policy by identifying best practices, lessons learned, or areas for improvement.

Advantages of Case Study Research

There are several advantages of case study research, including:

• In-depth exploration: Case study research allows for a detailed exploration and analysis of a specific phenomenon, issue, or problem in its real-life context. This can provide a comprehensive understanding of the case and its dynamics, which may not be possible through other research methods.
• Rich data: Case study research can generate rich and detailed data, including qualitative data such as interviews, observations, and documents. This can provide a nuanced understanding of the case and its complexity.
• Holistic perspective: Case study research allows for a holistic perspective of the case, taking into account the various factors, processes, and mechanisms that contribute to the case and its outcomes. This can help to develop a more accurate and comprehensive understanding of the case.
• Theory development: Case study research can help to develop and refine theories and concepts by providing empirical evidence and concrete examples of how they can be applied in real-life situations.
• Practical application: Case study research can inform practice or policy by identifying best practices, lessons learned, or areas for improvement.
• Contextualization: Case study research takes into account the specific context in which the case is situated, which can help to understand how the case is influenced by the social, cultural, and historical factors of its environment.

Limitations of Case Study Research

There are several limitations of case study research, including:

• Limited generalizability : Case studies are typically focused on a single case or a small number of cases, which limits the generalizability of the findings. The unique characteristics of the case may not be applicable to other contexts or populations, which may limit the external validity of the research.
• Biased sampling: Case studies may rely on purposive or convenience sampling, which can introduce bias into the sample selection process. This may limit the representativeness of the sample and the generalizability of the findings.
• Subjectivity: Case studies rely on the interpretation of the researcher, which can introduce subjectivity into the analysis. The researcher’s own biases, assumptions, and perspectives may influence the findings, which may limit the objectivity of the research.
• Limited control: Case studies are typically conducted in naturalistic settings, which limits the control that the researcher has over the environment and the variables being studied. This may limit the ability to establish causal relationships between variables.
• Time-consuming: Case studies can be time-consuming to conduct, as they typically involve a detailed exploration and analysis of a specific case. This may limit the feasibility of conducting multiple case studies or conducting case studies in a timely manner.
• Resource-intensive: Case studies may require significant resources, including time, funding, and expertise. This may limit the ability of researchers to conduct case studies in resource-constrained settings.

About the author

Muhammad Hassan

Researcher, Academic Writer, Web developer

You may also like

Questionnaire – Definition, Types, and Examples

Observational Research – Methods and Guide

Quantitative Research – Methods, Types and...

Qualitative Research Methods

Explanatory Research – Types, Methods, Guide

Survey Research – Types, Methods, Examples

Student Case Study

Delving into student case studies offers invaluable insights into educational methodologies and student behaviors. This guide, complete with detailed case study examples , is designed to help educators, researchers, and students understand the nuances of creating and analyzing case studies in an educational context. By exploring various case study examples, you will gain the tools and knowledge necessary to effectively interpret and apply these studies, enhancing both teaching and learning experiences in diverse academic settings.

What is a Student Case Study? – Meaning A student case study is an in-depth analysis of a student or a group of students to understand various educational, psychological, or social aspects. It involves collecting detailed information through observations, interviews, and reviewing records, to form a comprehensive picture. The goal of a case study analysis is to unravel the complexities of real-life situations that students encounter, making it a valuable tool in educational research. In a case study summary, key findings are presented, often leading to actionable insights. Educators and researchers use these studies to develop strategies for improving learning environments. Additionally, a case study essay allows students to demonstrate their understanding by discussing the analysis and implications of the case study, fostering critical thinking and analytical skills.

Download Student Case Study Bundle

Schools especially those that offers degree in medicine, law, public policy and public health teaches students to learn how to conduct a case study. Some students say they love case studies . For what reason? Case studies offer real world challenges. They help in preparing the students how to deal with their future careers. They are considered to be the vehicle for theories and concepts that enables you to be good at giving detailed discussions and even debates. Case studies are useful not just in the field of education, but also in adhering to the arising issues in business, politics and other organizations.

Student Case Study Format

Case Study Title : Clear and descriptive title reflecting the focus of the case study. Student’s Name : Name of the student the case study is about. Prepared by : Name of the person or group preparing the case study. School Name : Name of the school or educational institution. Date : Date of completion or submission.

Introduction

Background Information : Briefly describe the student’s background, including age, grade level, and relevant personal or academic history. Purpose of the Case Study : State the reason for conducting this case study, such as understanding a particular behavior, learning difficulty, or achievement.

Case Description

Situation or Challenge : Detail the specific situation, challenge, or condition that the student is facing. Observations and Evidence : Include observations from teachers, parents, or the students themselves, along with any relevant academic or behavioral records.

Problem Analysis : Analyze the situation or challenge, identifying potential causes or contributing factors. Impact on Learning : Discuss how the situation affects the student’s learning or behavior in school.

Intervention Strategies

Action Taken : Describe any interventions or strategies implemented to address the situation. This could include educational plans, counseling, or specific teaching strategies. Results of Intervention : Detail the outcome of these interventions, including any changes in the student’s behavior or academic performance.

Conclusion and Recommendations

Summary of Findings : Summarize the key insights gained from the case study. Recommendations : Offer suggestions for future actions or strategies to further support the student. This might include recommendations for teachers, parents, or the student themselves.

Best Example of Student Case Study

Overcoming Reading Challenges: A Case Study of Emily Clark, Grade 3 Prepared by: Laura Simmons, Special Education Teacher Sunset Elementary School Date: May 12, 2024   Emily Clark, an 8-year-old student in the third grade at Sunset Elementary School, has been facing significant challenges with reading and comprehension since the first grade. Known for her enthusiasm and creativity, Emily’s struggles with reading tasks have been persistent and noticeable. The primary purpose of this case study is to analyze Emily’s reading difficulties, implement targeted interventions, and assess their effectiveness.   Emily exhibits difficulty in decoding words, reading fluently, and understanding text, as observed by her teachers since first grade. Her reluctance to read aloud and frustration with reading tasks have been consistently noted. Assessments indicate that her reading level is significantly below the expected standard for her grade. Parental feedback has also highlighted Emily’s struggles with reading-related homework.   Analysis of Emily’s situation suggests a potential learning disability in reading, possibly dyslexia. This is evidenced by her consistent difficulty with word recognition and comprehension. These challenges have impacted not only her reading skills but also her confidence and participation in class activities, especially those involving reading.   To address these challenges, an individualized education plan (IEP) was developed. This included specialized reading instruction focusing on phonemic awareness and decoding skills, multisensory learning approaches, and regular sessions with a reading specialist. Over a period of six months, Emily demonstrated significant improvements. She engaged more confidently in reading activities, and her reading assessment scores showed notable progress.   In conclusion, the intervention strategies implemented for Emily have been effective. Her case highlights the importance of early identification and the implementation of tailored educational strategies for students with similar challenges. It is recommended that Emily continues to receive specialized instruction and regular monitoring. Adjustments to her IEP should be made as necessary to ensure ongoing progress. Additionally, fostering a positive reading environment at home is also recommended.

18+ Student Case Study Examples

1. student case study.

2. College Student Case Study

3. Student Case Study in the Classroom

Free Download

4. Student Case Study Format Template

• Google Docs

Size: 153 KB

5. Sample Student Case Study Example

stu.westga.edu

Size: 241 KB

6. Education Case Study Examples for Students

ceedar.education.ufl.edu

Size: 129 KB

7. Graduate Student Case Study Example

educate.bankstreet.edu

8. Student Profile Case Study Example

wholeschooling.net

Size: 51 KB

9. Short Student Case Study Example

files.eric.ed.gov

Size: 192 KB

10. High School Student Case Study Example

educationforatoz.com

Size: 135 KB

11. Student Research Case Study Example

Size: 67 KB

12. Classroom Case Study Examples

Size: 149 KB

13. Case Study of a Student

14. Sample Student Assignment Case Study Example

oise.utoronto.ca

Size: 43 KB

15. College Student Case Study Example

Size: 221 KB

16. Basic Student Case Study Example

Size: 206 KB

17. Free Student Impact Case Study Example

Size: 140 KB

18. Student Case Study in DOC Example

old.sjsu.edu

Size: 12 KB

19. Case Study Of a Student with Anxiety

Size: 178 KB

Case Study Definition

A case study is defined as a research methodology that allows you to conduct an intensive study about a particular person, group of people, community, or some unit in which the researcher could provide an in-depth data in relation to the variables. Case studies can examine a phenomena in the natural setting. This increases your ability to understand why the subjects act such. You may be able to describe how this method allows every researcher to take a specific topic to narrow it down making it into a manageable research question. The researcher gain an in-depth understanding about the subject matter through collecting qualitative research and quantitative research datasets about the phenomenon.

Benefits and Limitations of Case Studies

If a researcher is interested to study about a phenomenon, he or she will be assigned to a single-case study that will allow him or her to gain an understanding about the phenomenon. Multiple-case study would allow a researcher to understand the case as a group through comparing them based on the embedded similarities and differences. However, the volume of data in case studies will be difficult to organize and the process of analysis and strategies needs to be carefully decided upon. Reporting of findings could also be challenging at times especially when you are ought to follow for word limits.

Example of Case Study

Nurses’ pediatric pain management practices.

One of the authors of this paper (AT) has used a case study approach to explore nurses’ pediatric pain management practices. This involved collecting several datasets:

Observational data to gain a picture about actual pain management practices.

Questionnaire data about nurses’ knowledge about pediatric pain management practices and how well they felt they managed pain in children.

Questionnaire data about how critical nurses perceived pain management tasks to be.

These datasets were analyzed separately and then compared and demonstrated that nurses’ level of theoretical did not impact on the quality of their pain management practices. Nor did individual nurse’s perceptions of how critical a task was effect the likelihood of them carrying out this task in practice. There was also a difference in self-reported and observed practices; actual (observed) practices did not confirm to best practice guidelines, whereas self-reported practices tended to.

How do you Write a Case Study for Students?

1. choose an interesting and relevant topic:.

Select a topic that is relevant to your course and interesting to your audience. It should be specific and focused, allowing for in-depth analysis.

2. Conduct Thorough Research :

Gather information from reputable sources such as books, scholarly articles, interviews, and reliable websites. Ensure you have a good understanding of the topic before proceeding.

3. Identify the Problem or Research Question:

Clearly define the problem or research question your case study aims to address. Be specific about the issues you want to explore and analyze.

4. Introduce the Case:

Provide background information about the subject, including relevant historical, social, or organizational context. Explain why the case is important and what makes it unique.

5. Describe the Methods Used:

Explain the methods you used to collect data. This could include interviews, surveys, observations, or analysis of existing documents. Justify your choice of methods.

6. Present the Findings:

Present the data and findings in a clear and organized manner. Use charts, graphs, and tables if applicable. Include direct quotes from interviews or other sources to support your points.

7. Analytical Interpretation:

Analyze the data and discuss the patterns, trends, or relationships you observed. Relate your findings back to the research question. Use relevant theories or concepts to support your analysis.

8. Discuss Limitations:

Acknowledge any limitations in your study, such as constraints in data collection or research methods. Addressing limitations shows a critical awareness of your study’s scope.

9. Propose Solutions or Recommendations:

If your case study revolves around a problem, propose practical solutions or recommendations based on your analysis. Support your suggestions with evidence from your findings.

10. Write a Conclusion:

Summarize the key points of your case study. Restate the importance of the topic and your findings. Discuss the implications of your study for the broader field.

What are the objectives of a Student Case Study?

1. learning and understanding:.

• To deepen students’ understanding of a particular concept, theory, or topic within their field of study.
• To provide real-world context and practical applications for theoretical knowledge.

2. Problem-Solving Skills:

• To enhance students’ critical thinking and problem-solving abilities by analyzing complex issues or scenarios.
• To encourage students to apply their knowledge to real-life situations and develop solutions.

3. Research and Analysis:

• To develop research skills, including data collection, data analysis , and the ability to draw meaningful conclusions from information.
• To improve analytical skills in interpreting data and making evidence-based decisions.

4. Communication Skills:

• To improve written and oral communication skills by requiring students to present their findings in a clear, organized, and coherent manner.
• To enhance the ability to communicate complex ideas effectively to both academic and non-academic audiences.

5. Ethical Considerations:

To promote awareness of ethical issues related to research and decision-making, such as participant rights, privacy, and responsible conduct.

6. Interdisciplinary Learning:

To encourage cross-disciplinary or interdisciplinary thinking, allowing students to apply knowledge from multiple areas to address a problem or issue.

7. Professional Development:

• To prepare students for future careers by exposing them to real-world situations and challenges they may encounter in their chosen profession.
• To develop professional skills, such as teamwork, time management, and project management.

8. Reflection and Self-Assessment:

• To prompt students to reflect on their learning and evaluate their strengths and weaknesses in research and analysis.
• To foster self-assessment and a commitment to ongoing improvement.

9. Promoting Innovation:

• To inspire creativity and innovation in finding solutions to complex problems or challenges.
• To encourage students to think outside the box and explore new approaches.

10. Building a Portfolio:

To provide students with tangible evidence of their academic and problem-solving abilities that can be included in their academic or professional portfolios.

What are the Elements of a Case Study?

A case study typically includes an introduction, background information, presentation of the main issue or problem, analysis, solutions or interventions, and a conclusion. It often incorporates supporting data and references.

How Long is a Case Study?

The length of a case study can vary, but it generally ranges from 500 to 1500 words. This length allows for a detailed examination of the subject while maintaining conciseness and focus.

How Big Should a Case Study Be?

The size of a case study should be sufficient to comprehensively cover the topic, typically around 2 to 5 pages. This size allows for depth in analysis while remaining concise and readable.

What Makes a Good Case Study?

A good case study is clear, concise, and well-structured, focusing on a relevant and interesting issue. It should offer insightful analysis, practical solutions, and demonstrate real-world applications or implications.

Case studies bring people into the real world to allow themselves engage in different fields such as in business examples, politics, health related aspect where each individuals could find an avenue to make difficult decisions. It serves to provide framework for analysis and evaluation of the different societal issues. This is one of the best way to focus on what really matters, to discuss about issues and to know what can we do about it.

AI Generator

Text prompt

• Instructive
• Professional

Education Case Study Examples for Students

Graduate Student Case Study Example

Student Profile Case Study Example

High School Student Case Study Example

Student Research Case Study Example

Top 40 Most Popular Case Studies of 2021

Two cases about Hertz claimed top spots in 2021's Top 40 Most Popular Case Studies

Two cases on the uses of debt and equity at Hertz claimed top spots in the CRDT’s (Case Research and Development Team) 2021 top 40 review of cases.

Hertz (A) took the top spot. The case details the financial structure of the rental car company through the end of 2019. Hertz (B), which ranked third in CRDT’s list, describes the company’s struggles during the early part of the COVID pandemic and its eventual need to enter Chapter 11 bankruptcy. 

The success of the Hertz cases was unprecedented for the top 40 list. Usually, cases take a number of years to gain popularity, but the Hertz cases claimed top spots in their first year of release. Hertz (A) also became the first ‘cooked’ case to top the annual review, as all of the other winners had been web-based ‘raw’ cases.

Besides introducing students to the complicated financing required to maintain an enormous fleet of cars, the Hertz cases also expanded the diversity of case protagonists. Kathyrn Marinello was the CEO of Hertz during this period and the CFO, Jamere Jackson is black.

Sandwiched between the two Hertz cases, Coffee 2016, a perennial best seller, finished second. “Glory, Glory, Man United!” a case about an English football team’s IPO made a surprise move to number four.  Cases on search fund boards, the future of malls,  Norway’s Sovereign Wealth fund, Prodigy Finance, the Mayo Clinic, and Cadbury rounded out the top ten.

Other year-end data for 2021 showed:

• Online “raw” case usage remained steady as compared to 2020 with over 35K users from 170 countries and all 50 U.S. states interacting with 196 cases.
• Fifty four percent of raw case users came from outside the U.S..
• The Yale School of Management (SOM) case study directory pages received over 160K page views from 177 countries with approximately a third originating in India followed by the U.S. and the Philippines.
• Twenty-six of the cases in the list are raw cases.
• A third of the cases feature a woman protagonist.
• Orders for Yale SOM case studies increased by almost 50% compared to 2020.
• The top 40 cases were supervised by 19 different Yale SOM faculty members, several supervising multiple cases.

CRDT compiled the Top 40 list by combining data from its case store, Google Analytics, and other measures of interest and adoption.

All of this year’s Top 40 cases are available for purchase from the Yale Management Media store .

And the Top 40 cases studies of 2021 are:

1.   Hertz Global Holdings (A): Uses of Debt and Equity

2.   Coffee 2016

3.   Hertz Global Holdings (B): Uses of Debt and Equity 2020

4.   Glory, Glory Man United!

5.   Search Fund Company Boards: How CEOs Can Build Boards to Help Them Thrive

6.   The Future of Malls: Was Decline Inevitable?

7.   Strategy for Norway's Pension Fund Global

8.   Prodigy Finance

9.   Design at Mayo

10. Cadbury

11. City Hospital Emergency Room

13. Volkswagen

14. Marina Bay Sands

15. Shake Shack IPO

16. Mastercard

17. Netflix

18. Ant Financial

19. AXA: Creating the New CR Metrics

20. IBM Corporate Service Corps

21. Business Leadership in South Africa's 1994 Reforms

22. Alternative Meat Industry

23. Children's Premier

24. Khalil Tawil and Umi (A)

25. Palm Oil 2016

26. Teach For All: Designing a Global Network

27. What's Next? Search Fund Entrepreneurs Reflect on Life After Exit

28. Searching for a Search Fund Structure: A Student Takes a Tour of Various Options

30. Project Sammaan

31. Commonfund ESG

32. Polaroid

33. Connecticut Green Bank 2018: After the Raid

34. FieldFresh Foods

35. The Alibaba Group

36. 360 State Street: Real Options

37. Herman Miller

38. AgBiome

39. Nathan Cummings Foundation

40. Toyota 2010

Enhancing Procedural Writing Through Personalized Example Retrieval: A Case Study on Cooking Recipes

• Open access
• Published: 22 April 2024

Cite this article

You have full access to this open access article

• Paola Mejia-Domenzain   ORCID: orcid.org/0000-0003-1242-3134 1 ,
• Jibril Frej   ORCID: orcid.org/0009-0009-0631-0636 1 ,
• Seyed Parsa Neshaei   ORCID: orcid.org/0000-0002-4794-395X 1 ,
• Luca Mouchel 1 ,
• Tanya Nazaretsky   ORCID: orcid.org/0000-0003-1343-0627 1 ,
• Thiemo Wambsganss 1 ,
• Antoine Bosselut   ORCID: orcid.org/0000-0001-8968-9649 1 &
• Tanja Käser   ORCID: orcid.org/0000-0003-0672-0415 1  

Writing high-quality procedural texts is a challenging task for many learners. While example-based learning has shown promise as a feedback approach, a limitation arises when all learners receive the same content without considering their individual input or prior knowledge. Consequently, some learners struggle to grasp or relate to the feedback, finding it redundant and unhelpful. To address this issue, we present RELEX , an adaptive learning system designed to enhance procedural writing through personalized example-based learning. The core of our system is a multi-step example retrieval pipeline that selects a higher quality and contextually relevant example for each learner based on their unique input. We instantiate our system in the domain of cooking recipes. Specifically, we leverage a fine-tuned Large Language Model to predict the quality score of the learner’s cooking recipe. Using this score, we retrieve recipes with higher quality from a vast database of over 180,000 recipes. Next, we apply BM25 to select the semantically most similar recipe in real-time. Finally, we use domain knowledge and regular expressions to enrich the selected example recipe with personalized instructional explanations. We evaluate RELEX in a 2 x 2 controlled study (personalized vs. non-personalized examples, reflective prompts vs. none) with 200 participants. Our results show that providing tailored examples contributes to better writing performance and user experience.

Similar content being viewed by others

Design in Everyday Cooking: Challenges for Assisting with Menu Planning and Food Preparation

Taaable: A Case-Based System for Personalized Cooking

Accessibility-Driven Cooking System

Avoid common mistakes on your manuscript.

Introduction

Writing, decomposing, and revising texts are critical skills in many daily domains and professional environments. Procedural writing is a form of expository writing that promotes the replicability of procedures and the transfer of knowledge (Ambarwati & Listyani, 2021 ). Procedural texts are ubiquitous in many professions, examples include instruction manuals, algorithmic code (Ambarwati & Listyani, 2021 ), lab protocols, and cooking recipes (Alviana, 2019 ). Unfortunately, many learners struggle to write complete and high-quality procedural texts (Mejia-Domenzain et al., 2022 ; Ambarwati & Listyani, 2021 ).

Procedural writing is a so-called heuristic domain (Renkl et al., 2009 ), requiring a combination of knowledge of the learning domain (e.g., how to structure a procedural text) and the application domain (e.g., chemistry in the case of lab protocols). This domain dependence prevents the development of a single algorithmic solution for writing good procedural texts. In this context, learners can benefit from learning from examples. Learning from examples enables learners to "borrow" knowledge from others (Sweller, 1994 ) and abstract general rules that can be used to solve similar problems in the future. Prior research has mainly focused on example-based learning applied to highly structured tasks like mathematics and physics (Sweller, 1994 ; Hilbert et al., 2008 ; van Gog et al., 2008 ). Nevertheless, example-based learning has been studied in heuristic domains with no single correct solution (Renkl et al., 2009 ). In these contexts, the examples are often enriched to include instructional explanations that can reduce the cognitive load by emphasizing relevant characteristics (Schworm & Renkl, 2007 ; van Gog et al., 2008 ). However, the provided examples and instructional explanations are commonly static (Renkl, 2002 ): all learners are provided with the exact same content (e.g., a worked-example by an expert with instructional explanation), independent of their actual skill level. Hence, the provided examples and instructions might be too complex or not relevant to the user, hindering learning and motivation (van Gog et al., 2008 ; Alamri et al., 2020 ).

Providing tailored examples and feedback timely, therefore, has the potential to increase learner performance and experience. While there exists a large body of research on optimal task selection in structured domains (e.g., Bassen et al. ( 2020 )), only a few works have focused on retrieving examples tailored to the user’s context in heuristic domains. Existing research has, for example, employed feature-based similarity metrics (Hosseini & Brusilovsky, 2017 ; Pelánek, 2020 ) or unsupervised semantic sentence similarity methods (Zlabinger et al., 2020 ) to retrieve similar educational items. However, the majority of these works focused on retrieving similar (in terms of the input text provided by the user) expert-created examples, disregarding the actual skill level of the user.

Furthermore, there is also a vast research on providing personalized explanations and instructions for various writing tasks. Existing tools visualize the revision history of the user’s text (Afrin et al., 2021 ) or use an underlying domain-specific structure to enrich the user’s text with feedback and explanations (Wang et al., 2020 ). However, they do not provide suggestions or examples on how to correct the shortcomings in the user’s text.

In this paper, we present RELEX (REcipe Learning through EXamples), an effective and scalable learning system for procedural writing using personalized example-based learning. We have instantiated RELEX in the domain of cooking recipes because of its familiarity and practical relevance to culinary students and chef apprentices, as identified by prior work (Mejia-Domenzain et al., 2022 ). RELEX features a multi-step pipeline retrieving an example that is 1) relevant for the learner (i.e., similar in terms of topic), 2) of better quality than the learner’s text (i.e., tailored to the learner’s skill level), and 3) annotated with explanations and suggestions that the learner’s text is lacking. Our pipeline takes as input the learner’s recipe and predicts its quality using a fine-tuned Large Language Model (LLM). Then, it retrieves a set of texts with a higher quality (than the predicted quality) from a database containing over \(180'000\) rated recipes. Finally, the most semantically similar recipe is extracted from the retrieved candidate set using BM25 .

To evaluate RELEX , we conduct a \(2\times 2\) controlled study with 200 participants, in which we manipulate a) the adaptiveness of the provided example and annotations (adaptive vs. non-adaptive example and feedback), and b) the prompts for reflection (reflective prompts vs. none). We also run the same task with a control group receiving static procedural writing support only. With our analyses, we aim to address the following three research questions: What are the effects of providing a personalized example along with adaptive feedback and reflective guidance on learners’ experience (RQ1), writing performance (RQ2) and revising behavior (RQ3)?

Our results indicate that participants who received tailored examples revised their cooking recipes more, wrote them with higher quality, and had a more positive perception of the tool than the users without adaptive feedback.

Related Work and Conceptual Background

In this paper, we present the design and evaluation of a learning system for personalized example-based learning at scale, which is instantiated in the domain of procedural writing. Our study has therefore been influenced by related work in the areas of (1) learning procedural writing skills, (2) example-based learning in heuristic domains, and (3) adaptive learning.

Learning Procedural Writing Skills

Procedural writing, a form of expository writing, facilitates the transfer of knowledge and the replicability of procedures (Ambarwati & Listyani, 2021 ). This type of writing finds its applications in various fields, ranging from life sciences lab protocols to technical documentation and culinary recipes (Wieringa & Farkas, 1991 ; Mejia-Domenzain et al., 2022 ; Alviana, 2019 ).

While procedural writing is highly dependent on the subject matter, previous research (Wieringa & Farkas, 1991 ; Sato & Matsushima, 2006 ; Traga Philippakos, 2019 ; Adoniou, 2013 ) has identified three main qualities of high-quality procedural texts: structure, clarity, and specificity. Structure refers to the organization of the text like having appropriate sections. Clarity involves providing necessary details, and specificity refers to the use of appropriate, domain-specific vocabulary.

Previous research has found that learners often encounter difficulties when attempting to compose comprehensive and high-quality procedural texts (Mejia-Domenzain et al., 2022 ; Ambarwati & Listyani, 2021 ). Common mistakes, in the case of computer documentation and nuclear power plants procedures, are the incorrect order of steps, missing elements, lack of details, or ambiguous words that lead to confusion (Wieringa & Farkas, 1991 ). Similarly, the recipes documented by chef apprentices are often missing ingredients and exhibit a lack of detail and use of specific vocabulary (Mejia-Domenzain et al., 2022 ).

Given these challenges in writing procedural texts, the question arises: How can we effectively teach and instruct this skill? Effective feedback mechanisms for procedural writing have received limited attention. One notable investigated mechanism involved feedback through simulation: students were prompted to compose a procedural text detailing how to draw a geometrical figure and subsequently received feedback in the form of the figure drawn based on their instructions (Sato & Matsushima, 2006 ).

While there are general learning objectives (structure, clarity, and specificity), the dependence on the domain prevents the development of a single algorithmic solution for writing a good procedural text. In this context, learners can benefit from learning from examples. Previous research has investigated the efficacy of model-based instruction, where students observe a teacher demonstrating and verbally describing the procedure in action. Notably, studies have applied this approach in various scenarios, such as making a peanut butter and jelly sandwich (Traga Philippakos, 2019 ) and preparing a chicken sandwich (Alviana, 2019 ). Encouragingly, both works reported positive effects on the quality of procedural writing resulting from the implementation of the demonstration technique. Surprisingly, despite the proven benefits of using written worked examples in other genres, such as argumentation skills (Schworm & Renkl, 2007 ), their potential application in procedural writing remains largely unexplored.

Example-Based Learning in Heuristic Domains

Example-based learning is an effective method to acquire knowledge by observing and/or imitating what other people do, say, or write (Sweller, 1994 ). It allows learners to build a cognitive schema of how problems should be solved. In addition, learners can abstract general rules from the examples and ultimately transfer and adapt them to other problems (van Gog & Rummel, 2010 ). The vast majority of research on example-based learning has studied their effectiveness in well-structured tasks, such as algebra (Sweller, 1994 ) and physics (van Gog et al., 2008 ). More recently, worked-examples and solved-examples have been applied to non-algorithmic learning domains such as argumentative writing (Schworm & Renkl, 2007 ) and mathematical proof finding (Hilbert et al., 2008 ). In heuristic domains (Renkl et al., 2009 ), where no algorithmic solution can be provided (e.g. cooking recipes), learners acquire heuristics that help them find a solution. Examples in heuristic domains require learners to process two different content levels: (1) the learning domain (i.e., how to structure the solution) and (2) the exemplifying domain (i.e., the topic). In the case of cooking recipes, learners need to understand how to structure a procedural text (learning domain: procedural writing) and be familiar with the cooking domain (the exemplifying domain). Given the two content levels, these examples are referred to as double-content . In structural domains, worked-examples are usually annotated with the steps to solve the problem. In contrast, the double-content examples tend to be enriched with self-explanation prompts and/or additional instructional explanations.

Reflective Prompts . According to the self-explanation effect , learners benefit more from the examples if they can actively explain the examples to themselves (Wong et al., 2002 ). Furthermore, the quality of the self-explanations determines what is learned from the examples (Chi et al., 1989 ). However, frequently, learners’ self-explanations are superficial or passive. Thus, the application of prompts is a possible intervention to increase the quality and depth of the explanations. These prompts should stimulate the active processing of learning materials and direct attention to the central aspects (Schworm & Renkl, 2007 ). The use of self-regulated learning (SRL) prompts has been shown to foster conceptual knowledge (Roelle et al., 2012 ). Furthermore, SRL prompts (i.e., which aspects of the learning materials do you find interesting, useful, and convincing, and which not? ) have been used to help the learner focus on the central elements of examples (Nückles et al., 2009 ) or to guide learners to diagnose their deficiencies and be critical (Fan et al., 2017 ).

Instructional Explanations . Instructional explanations are another possibility to enrich examples. It has been demonstrated that in a first learning phase, instructional explanations improve the learning outcomes compared to when there are no explanations provided  (van Gog et al., 2008 ). However, these explanations can be detrimental later in the learning, since the provided information soon becomes redundant and the explanations increase the cognitive load and hinder learning. Instructional explanations have the following disadvantages in comparison to self-explanation (Renkl, 2002 ): (1) they are not adapted to the learner’s prior knowledge, so they can be redundant or too complex and hard to understand; (2) they are often not timely and therefore hard to integrate as part of the ongoing learner’s activities.

In a \(2\times 2\) study on the effect of self-explanation prompts and instructional explanations, the group that received only self-explanation prompts had the most favorable learning outcomes, whereas the group that received instructional explanations had the highest perception of learning (Schworm & Renkl, 2007 ). Nevertheless, the authors did not examine the use of adaptive instructional explanations. A first step in this direction has been taken by providing so-called faded examples in geometry learning (Schwonke et al., 2009 ). Students were shown complete worked-out examples at first; over time steps from the example were gradually removed. However, the missing steps and the selected examples were pre-determined and not chosen adaptively depending on the students.

To summarize, the provided examples, the reflective prompts, and the instructional explanations are commonly static: all learners are provided with the same content (e.g., a worked-example by an expert with instructional explanation). The examples and explanations are (1) not adapted to the learner’s prior knowledge, so they can be redundant and hence hinder learning (van Gog et al., 2008 ) and (2) not timely and relevant, hence decreasing engagement (Alamri et al., 2020 ). Providing personalized examples and instruction in a timely manner therefore has the potential to improve learning.

Adaptive Learning

Providing personalized examples and adaptive annotations and explanations translates into providing 1) personalized content (the example) and 2) personalized instruction.

Personalized Content . In content level adaptation, the learning objects (e.g., examples, tasks) are selected and adapted based on the content (e.g., current task, answer, knowledge state) of the user (Premlatha & Geetha, 2015 ). One approach to providing personalized content is to retrieve a tailored example from an existing collection. The collection consists of all the examples available, the query is the user’s context and the system ranks examples in the collection based on their similarity with the user’s context. Depending on the task to be learned, the user’s context can be the current task, the answer, the learner’s knowledge or any combination of these. Example retrieval involves three steps: (1) computing a similarity between the learner’s context and examples from the collection, (2) ranking the examples based on their similarity and (3) presenting the most similar or top- k examples to the learner. For instance, Hosseini and Brusilovsky ( 2017 ) used semantic-level similarity-based linking to recommend personalized examples to programming learners.  Pelánek ( 2020 ) explored feature-based (such as the occurrence of domain-specific keywords) and performance-based measures to compare the similarity of educational items in various domains. Furthermore,  Zlabinger et al. ( 2020 ) provided crowdworkers with personalized examples: they used unsupervised semantic sentence similarity methods to retrieve tailored expert-labeled examples.

Obtaining high-quality expert examples for learning purposes can be challenging and costly. In such cases, peer examples serve as an alternative, which, despite their potential loss in quality, can prove more effective in a learning scenario (Doroudi et al., 2016 ). However, evaluating the quality of peer examples poses its own challenge, as the perception of good quality varies among raters, tasks, and genres (Wilson et al., 2014 ). To address this issue, recent research has explored the application of LLMs, like BERT (Devlin et al., 2019 ) or GPT-models (Brown et al., 2020 ), for tasks such as automatically scoring essays (Mayfield & Black, 2020 ), rating recipe nutritional quality (Hu et al., 2022 ), and evaluating text generation (Sellam et al., 2020 ). These LLMs, being at the forefront of natural language processing (NLP) tasks (Devlin et al., 2019 ; Liu et al., 2019 ; Brown et al., 2020 ), offer a promising approach to predict the quality of examples in heuristic domains.

Personalized Instruction . In contrast to generic instruction, personalized instruction (or feedback or explanation) is dynamic, which means that different learners will receive different information (Bimba et al., 2017 ). While there is a range of research on providing personalized feedback and hints in structured domains such as mathematics (Paassen et al., 2018 ) or programming (Ahmed et al., 2020 ), less work has focused on giving automated fine-grained suggestions and explanations in heuristic domains such as expository writing.

Existing NLP-based writing support tools often provide holistic feedback on higher-level properties of the text such as grammar errors, fluency, or coherence (e.g., Grammarly  (Max et al., 2022 )). To provide more detailed guidance, other tools adopt alternative approaches. For instance, ArgRewrite (Afrin et al., 2021 ) visualizes revision histories by annotating a side-by-side comparison of two drafts, providing revision suggestions at the sentence and sub-sentence level. In contrast, ArguLens (Wang et al., 2020 ) utilizes a domain-specific structure by imposing an argumentation-enhanced representation, breaking the user text into argumentation components and standpoints. Despite these valuable contributions, none of the existing systems combine adaptive instruction with a comparison example that could leverage the potential of example-based learning.

RELEX - Learning With Personalized Examples

To study the effect of personalized example-based feedback on learners’ writing performance, revision behavior, and learning experience, we designed RELEX (REcipe Learning through EXamples). The primary purpose of RELEX is to facilitate procedural writing by providing students with tailored examples, accompanied by relevant annotations and reflective prompts. The tool aims to address three key aspects of procedural writing: (a) the organization and structure of texts, (b) the provision of requisite details for enhanced clarity, and (c) the appropriate utilization of specific vocabulary. In the following, we will describe the two main components of RELEX , the user interface and the personalized example retrieval pipeline.

Interaction flow: A learner requests feedback (F1) and receives a tailored example recipe (F2) with highlighted in-text elements (F3) and personalized explanations (F4). The learner is also prompted to reflect on the strengths and weaknesses of the recipes (F5)

User Interface

The user interface of RELEX is illustrated in Fig. 1 Footnote 1 . The main interface is shown on the upper part of the figure. The interface is split into two main panels: the Text Editor (left) where learners can write or edit recipes and request feedback by clicking "Analyze" ( \(F_1\) ) and the Personal Dashboard (right) displaying a selected example recipe with personalized annotations. This Personal Dashboard is again split vertically into two sections, listing suggestions to improve the recipe on the left ( \(F_4\) ) and showing the example recipe ( \(F_2\) ) with missing aspects in the learner’s recipe highlighted ( \(F_3\) ) on the right. Below the main interface (Fig. 1 bottom right), other types of recipe improvement tips together with a fragment of the example recipe that fulfills these suggestions are shown. More specifically, the bottom middle panel shows examples of tips on the specificity of ingredients and steps; and to the right, there are examples on the clarity of the steps. Finally, the Reflection Space ( \(F_5\) , bottom left) invites the user to carefully study the example recipe with the question What aspects of this example recipe do you find interesting, useful, convincing, and which not? ; and compare it to their own recipe with the question What deficiencies does your recipe have compared to the example recipe on the right? . The (synthetic) example in Fig.  1 illustrates these design functionalities. The learner has asked for feedback ( \(F_1\) ) on a recipe including chicken and is provided a similar recipe ("Louisiana Chicken") of higher quality (immediately visible by its clear structure) as an example ( \(F_2\) ). The highlights indicate the missing structural elements (for example, "List each ingredient separately", \(F_3\) ) and the left-hand pane of the personal dashboard suggests other tips on the structure like enumerating the steps ( \(F_4\) ). Other examples of \(F_3\) and \(F_4\) are shown below the main interface. In addition, the reflection panel ( \(F_5\) ) opens to the bottom left where the user answers the questions. The five design functionalities of RELEX (see Table 1 ) are based on design requirements derived from user interviews as well as from literature.

User Requirements . Given that the users should be the main focus of a design effort (Cooper et al., 2007 ), we conducted ten semi-structured user interviews (female-identifying: 6, male-identifying: 4) to better understand the specific user needs when using example-based learning in the context of procedural writing Footnote 2 . Participants described their past experiences with writing procedural texts, which included tutorials, lab protocols, technical manuals, and cooking recipes. One common difficulty they encountered when writing procedural texts was being too vague, missing steps, and having the readers struggle to reproduce the instructions they had written.

From these semi-structured interviews, we derived 22 user stories. The stories contained a multitude of detailed suggestions, such as the type of colors used for highlighting elements of the text or the request to see explanations for the highlighted elements. We clustered the different user stories based on the underlying topic and obtained five groups, from which we formed the following user requirements:

Examples should be relevant and similar so that the users can relate to them.

Users should be able to see more than one example in order to generalize and abstract the relevant elements.

The important elements of the text should be highlighted with different colors (indicating what each color means) to stimulate active processing.

The mechanism should have interactive explanations (e.g., when the mouse scrolls on top or clicks) of the highlighted text in the form of suggestions or questions (that can be dismissed) to help learners understand the underlying structure of the example.

The mechanism should include self-explanation and self-reflection prompts to foster active understanding of the example.

Literature Requirements . After deriving the user-centric requirements, we complemented them with the large body of literature on example-based learning (described in detail in “ Example-Based Learning in Heuristic Domains ”). The impact of this approach is highly dependent on the design of the examples utilized. With this regard, previous research examined various design aspects such as self-explanation prompts (Schworm & Renkl, 2007 ), content guidance (Renkl et al., 2009 ), and highlighting (Ringenberg & VanLehn, 2006 ). In their review paper, van Gog and Rummel ( 2010 ) synthesized these aspects and provide design guidelines for example processing. Similarly, Renkl ( 2002 ) derived design principles for instructional explanations. Drawing from the insights of these two review papers, we establish the literature-based design requirements of RELEX :

Active processing of examples should be stimulated by emphasizing important aspects of the procedure. This will help learners understand the underlying structure and transfer that knowledge to a different task (van Gog & Rummel, 2010 ).

Learners should be instructed to self-explain the example in order to foster active processing and understanding (van Gog & Rummel, 2010 ).

Examples and explanations should be presented on learners’ demand to ensure that they are appropriately timed and used in ongoing knowledge-building activities (Renkl, 2002 ).

Explanations should be short and minimalist to reduce the effort to process them (Renkl, 2002 ).

Explanations should not tell learners things that they already know or do not need to know (Renkl, 2002 ).

Table 1 illustrates the relationship between user and literature requirements and the corresponding functionalities of the tool. The design of these functionalities focused on meeting both the needs identified in the relevant literature and those expressed by users, with the goal of creating a tool that is both educationally effective and user-friendly. Specifically, we began by considering user requirements and then incorporated requirements derived from the literature where applicable. For instance, \(F_1\) caters to the users’ need for accessing multiple examples ( \(U_2\) ) and also aligns with the literature’s emphasis on the availability of on-demand examples ( \(L_3\) ). Similarly, \(F_3\) fulfills the requirement of highlighting important aspects of the learning material ( \(L_1\) ) by using different colors, a feature specifically requested by users ( \(U_3\) ), while also ensuring that redundancies are minimized ( \(L_5\) ). Furthermore, \(F_4\) supports the users’ desire for explanations ( \(U_4\) ) and the use of varied colors ( \(U_3\) ), while also adhering to the recommendation for brevity and minimalism in explanations ( \(L_4\) ). Additionally, \(F_5\) addresses the users’ preference for self-explanation prompts ( \(U_5\) ) in line with literature insights ( \(L_2\) ). Finally, \(F_2\) , which responds to the users’ desire for relevant and similar examples, represents an innovative aspect of our work.

Personalized Example Retrieval Pipeline

To retrieve tailored examples for learners, we propose a multi-step recipe selection pipeline. Our pipeline retrieves a personalized example recipe that satisfies the following constraints: 1) describing a similar dish, thus relevant to the learner, 2) of higher quality than the learner’s recipe, 3) annotated with explanations and suggestions based on identified weaknesses of the learner’s recipe, and 4) retrieved in real-time.

Hence, both the retrieved example and the highlighted suggestions are tailored to the learner’s content (e.g., the type of recipe) and skill level (e.g., the quality of the recipe). The pipeline is illustrated in Fig.  2 . It features an offline and an online component. The offline component (top, in green) consists of three main steps:

Preprocessing: a large cooking recipe database ( RecipeNLG ) is preprocessed to obtain the ratings for each recipe. We denote the resulting dataset of rated recipes as RELEXset .

Fine-Tuning: a general domain language model is fine-tuned on all recipes from RecipeNLG . This model is further fine-tuned on the regression task to predict the stars from RELEXset . We call this fine-tuned model RELEXset-predictor .

Recipe Annotation: the recipies from RELEXset are annotated using writing suggestions obtained from experts. We denote the resulting dataset of annotated, rated recipes as RELEX-sugg-set .

The online component (orange, Fig.  2 bottom), processes the learner’s recipe x in four steps:

Quality Prediction: the stars (quality) of x , denoted as S ( x ), is predicted using RELEXset-predictor .

Recipe Retrieval: all recipes of higher quality than x ( \(SB_x\) ) are retrieved from RELEX-sugg-set .

Relevance Filtering: only relevant recipes according to the missing suggestions are kept. We denote this filtered set as \(rel(SB_x)\) .

Recipe Similarity: recipe similarity is computed to output the most similar example recipe from \(rel(SB_x)\) .

In the following subsections, we describe each step of the offline and online components in detail.

Offline Training and Annotation

As seen in Fig. 2 , the offline phase consists of three steps: Preprocessing the database, training the cooking domain LLM for rating prediction, and annotating the recipes with improvement suggestions. Specifically, we quantify the quality of the recipes using crowd-sourced ratings, which allows us to sort the recipes based on the community’s perception. Then, we train a model to predict the rating (in the form of stars) a new recipe x would obtain, enabling us to extract recipes of higher quality than the recipe x from the database.

Preprocessing . We use RELEXset , a database composed of rated cooking recipes Footnote 3 . The recipes were extracted from RecipeNLG  (Bień et al., 2020 ), a publicly available dataset of clean and formatted versions of cooking recipes. Ratings are real numbers from 0 to 5. They were obtained from food.com , an online recipes site (Majumder et al., 2019 ). We remove from RELEXset all recipes with no ratings. As a result, RELEXset contains more than 180, 000 clean and formatted recipes with more than 700, 000 user ratings. One common problem with user ratings is that different users adopt different criteria and rating scales. Some users might, for example, be more tolerant than others and give higher ratings in general (Jin & Si, 2004 ). To mitigate this bias, following common practices in collaborative filtering models (Jin & Si, 2004 ), we standardize the ratings per user. We denote by \(R_y(x)\) the rating of user y for recipe x and by \(\hat{R}_y\) the average rating of user y across all recipes. Standardization consists in centering \(R_y(x)\) around \(\hat{R}_y\) with a unit standard deviation as follows: \(\widehat{R}_y\left( x\right) = \left( R_y\left( x\right) - \overline{R}_y\right) \bigg /\sqrt{\sum _{z \in \mathcal {X}} \frac{1}{|\mathcal {X}|} \left( R_y(z) - \overline{R}_y \right) ^2}\) with \(\mathcal {X}\) the set of all recipes in RELEXset . As the standardized rating cannot be computed when the standard deviation is 0, users who have only rated one recipe are automatically excluded from the analysis. To obtain a unique rating S ( x ) associated with each recipe, we average the standardized ratings across all users: \(S(x) = \sum _{y \in \mathcal {Y} }\widehat{R}_y\left( x\right) \big /|\mathcal {Y}|\) with \(\mathcal {Y}\) the set of all users in RELEXset . In the remaining part of the paper, we use “stars” to refer to the averaged user-standardized ratings.

Fine-Tuning on Recipes . Given that the recipes consist of text, we follow the recent advances in NLP (Devlin et al., 2019 ; Liu et al., 2019 ; Sanh et al., 2019 ; Brown et al., 2020 ) and use a pre-trained LLM to predict the quality (starts) of a recipe. The choice of pre-trained LLM is based on performance and efficiency. On the one hand, BERT, a widely-recognized LLM, employs self-attention mechanisms to generate context-aware word representations (Devlin et al., 2019 ). While BERT offers robust performance, RoBERTa, an enhanced version, surpasses it in various NLP benchmarks due to extensive training and hyperparameter optimization (Liu et al., 2019 ). On the other hand, RoBERTa’s computational demands are substantial, making it less ideal for real-time applications. To balance performance and efficiency, we opt for DistilRoBERTa, a streamlined version of RoBERTa (Sanh et al., 2019 ). Developed through knowledge distillation, DistilRoBERTa maintains much of RoBERTa’s efficacy but with reduced complexity, making it more suitable for our requirement of real-time prediction without relying on GPUs. This is in line with studies suggesting that increased prediction time can negatively impact user experience (Nah, 2003 ). Therefore, we initialize our predictor with the distilroberta-base checkpoint from HuggingFace’s transformers (Wolf et al., 2019 ).

It is worth noting that, distilroberta-base was trained on general texts from the internet and not specifically in the cooking domain. Following common practices (Gururangan et al., 2020 ; Sun et al., 2019 ), before fine-tuning the model for rating recipes, we first adapt distilroberta-base to the cooking domain by fine-tuning it on a Masked Language Modeling (MLM) task on the entire set of recipes from RecipeNLG . We will refer to the resulting model as RELEXset-MLM .

Fine-Tuning on a Regression Task . Given that we want to predict the averaged user-standardized rating (stars) of a recipe, we formulate the prediction stage as a regression task: for any given recipe denoted as x , the predictive model should output a real-valued star rating symbolized as S ( x ). Thus, we fine-tune RELEXset-MLM to predict the number of stars of recipes in RELEXset . We will refer to the obtained model as RELEXset-Predictor Footnote 4 . The model has six transformer layers, each with a hidden size of 768, and employs 12 attention heads. The intermediate layers in the transformers have a size of 3072. Moreover, the model uses GELU as its activation function and dropout rates for both attention probabilities and hidden layers are set to 0.1 Footnote 5 . Following, we use a fully connected neural network with one hidden layer that takes the [CLS] token final embedding as input and outputs the number of stars S ( x ). We optimize both RELEXset-MLM and RELEXset-Predictor using adam  (Kingma & Ba, 2015 ) with early stopping. Both RecipeNLG and RELEXset are split into train/validation/test sets with a ratio of 80/10/10. This ratio was chosen to provide sufficient data for training while also allowing adequate samples for validation and testing. Given the complexity of the model, the 80/10/10 split ensures that more data is available for training. Furthermore, given the large size of the dataset, \(10\%\) of the data points used for validation and testing are sufficient to validate and test effectively. We used the Kolmogorov-Smirnov test Footnote 6 , a nonparametric test of the equality of continuous probability distributions, to verify that there were no significant differences (train vs validation: \(p=.36\) , train vs test: .91, validation vs test: \(p=.75\) ) between the label distributions in the train, validation and test sets Footnote 7 . Learning rate, batch size, and weight decay were selected on the validation set using grid search from {1e-6, 1e-5, 2e-5, 3e-5, 5e-5, 1e-4}, {32, 64, 128, 256, 512} and {0.01, 0.02, 0.03, 0.05, 0.08, 0.1} respectively. We chose the best model (hereafter referred as RELEXset-Predictor ) based on the validation loss and tested its performance on the hold-out test set. RELEXset-Predictor achieved a mean absolute error (MAE) of 0.39 on the test set, which is slightly better than the baseline MAE of 0.42 (simply predicting the mean). Despite the difference not being significant, RELEXset-Predictor has the ability to generalize to new, unseen data, making it a more reliable tool for making predictions in real-world scenarios than the static baseline predictor. As outlined in “ Online Prediction and Selection ”, the subsequent stages of the pipeline are designed to address the prediction uncertainties by selecting recipes that fall within a quality range set above the MAE threshold to ensure that the recipe is perceived as better by the users.

Recipe Annotation . After choosing a targeted example recipe, we enrich the recipe with suggestions on how to improve the text. These suggestions are based on the three main aspects of high-quality procedural text (Wieringa & Farkas, 1991 ; Sato & Matsushima, 2006 ; Traga Philippakos, 2019 ): structure (i.e., clear organization of the text), clarity (i.e., appropriate degree of detail), and specificity (i.e., proper use of technical terms). The suggestions can be divided into general suggestions concerning the learning domain (i.e., how to write procedural text) and into suggestions specific to the exemplifying domain (cooking recipes). The domain-general suggestions are derived from the main qualities of good procedural text identified in previous work (Wieringa & Farkas, 1991 ; Sato & Matsushima, 2006 ; Traga Philippakos, 2019 ). The domain-specific suggestions are derived from "The Recipe Writer’s Handbook, Revised and Expanded" (Ostmann & Baker, 2001 ). In this handbook, two recipe book editors give punctual recommendations on how to write a good recipe in terms of the learning objectives (structure, specificity, and clarity). We use the keywords “specify” and “indicate” to retrieve 45 suggestions from the handbook. Table 2 lists all the domain-general suggestions as well as examples of domain-specific suggestions. There are four suggestions related to the structure and three suggestions related to the clarity of the text. For these two categories, there is a direct mapping between domain-general and domain-specific annotations. There are in total 38 recipe-specific suggestions related to the specificity of the steps and material Footnote 8 .

We transform the suggestions into explicit rules to be able to annotate each recipe for each of the 45 suggestions. Specifically, we classify each of the 45 suggestions as "followed", "missing", or "not relevant" for each recipe. For example, if the recipe does not require a pan, the suggestion to “ indicate the size and type of the pan ” is not relevant; on the other hand, if the recipe requires a “pan”, but the size (small, medium, large) or type (frying, skillet, non-stick, ceramic, etc) are not specified, the suggestion is “missing”. To facilitate this classification, we employ a rule-based system using regular expressions. This method allows for an automated annotation of the recipes. Our classification algorithm operates in two stages. Initially, it scans the recipe for keywords related to each suggestion (main keyword). Following the example, it would look for “pan” or synonyms. Subsequently, when a keyword is identified, the algorithm examines a 20-character range surrounding it to detect any mention of the specific characteristics detailed in the suggestion (supporting keywords). In our example, it would look for the size or type of the pan. This process is repeated for all suggestions, and the results are compiled into a dictionary. This dictionary reflects the status of each suggestion (followed, missing, or not relevant) for every recipe, including the specific locations where these criteria are met.

The previously described classification algorithm aims at ascertaining the presence of the specified keywords (supporting keywords) in proximity to another predetermined keyword (main keyword). We define “proximity” as a 20-character range to account for intervening descriptors (such as adjectives or qualifiers) that are typically positioned close to their corresponding nouns that might not be related to the suggestion. For example, for the suggestion about specifying the form of nuts (e.g., whole, halved, chopped, etc) in proximity to a nut’s name (e.g., walnut, almonds), a phrase like “slivered (form) blanched almonds (nut)” exemplifies a case where looking at the preceding or succeeding word fails to recognize the relationship due to the intervening descriptors. Given that the average word length in English is 4.8 Footnote 9 , we chose a 20-character range that is approximately 4 words apart. Empirical trials confirmed that this range effectively captures the necessary details in the majority of cases, striking a good balance between capturing essential information and excluding unrelated text that might pertain to other ingredients or elements rather than describing the main keyword.

To assess the rule-based annotation performance, we conducted an evaluation using a random sample of recipe segments. Two annotators, who are also authors of this work, were involved in this process. One of the annotators had participated in the generation of the rule-based regular expressions, while the other annotator was unfamiliar with the generation process. The choice of annotators was a pragmatic decision that allowed us to evaluate the rule-based model without the need for recruitment of external annotators. Per each suggestion, we randomly selected five recipe segments where the two-step annotation algorithm indicated that the suggestion was present and five segments where it was missing. The segments were shuffled and manually labeled to indicate whether the rule was being fulfilled or not. The Cohen’s Kappa score between the annotators was 0.85 (near perfect agreement (Landis & Koch, 1977 )) and the average accuracy was 0.95. We acknowledge that the choice of annotators could have induced a level of subjective interpretation. However, the random selection and shuffling of segments for annotation likely mitigated any subconscious biases. Moreover, the high inter-rater reliability indicates that the suggestions provided were clear and consistent, regardless of the annotators’ prior involvement in the process.

Online Prediction and Selection

The online part of the pipeline consists of retrieving a tailored comparison recipe for the user in real-time.

Quality Prediction . When a participant y asks for feedback on a recipe x , the first step consists in predicting the stars of the input recipe \(S_y(x)\) using RELEXset-Predictor .

Recipe Retrieval . In the next step, a candidate subset \(SB_x\) of recipes with higher quality (i.e., a higher stars value) is retrieved from RELEX-sugg-set . \(SB_x\) contains all the recipes with a rating withing the range \([S_y(x) + 0.4, S_y(x) + 0.8]\) . For example if the rating of the input recipe \(S_y(x)=1\) , \(SB_x\) will contain all the recipes with a standardized rating within the range [1.4, 1.8]. This range was chosen based on RELEXset-Predictor ’s MAE (0.39) as we did not want \(SB_x\) to contain recipes that fit within the error range of the predictor. Moreover, we wanted to show the user a peer recipe that is of better quality, but still similar enough for the user to relate to and not be discouraged by peer excellence (Rogers & Feller, 2016 ). We tested the selected range in a pilot study with 10 participants. We asked the participants to evaluate the level of the recipes seen in comparison to theirs, and the options were "much worse", "worse", "same level", "better" and "much better". None of the participants stated that the recipes were "much better", \(60\%\) perceived the recipe as better, and \(40\%\) as their same level.

Relevance Filtering . The next stage of the pipeline consists of filtering the candidate subset \(SB_x\) according to relevance. We consider that a recipe contains relevant feedback if it can exemplify how to successfully improve the input recipe x . To assess the relevance of the candidate recipe, we first identify the suggestions that are missing from the input recipe x . We then filter out from \(SB_x\) the recipes that do not contain relevant feedback. We postulate that a recipe contains relevant feedback if it follows at least one suggestion that is missing from x . We denote as \(rel\left( SB_x\right) \) the set of recipes from \(SB_x\) containing relevant feedback. To exemplify the filtering stage, let us consider the following minimal example: z is a recipe where the only suggestion classified as missing is " indicate the intensity of the heat ". Therefore, we will remove from \(SB_z\) all recipes that do not specify the intensity of the heat when they should have. Thus, the resulting set \(rel\left( SB_z\right) \) will contain only recipes that follow the suggestions: " indicate the intensity of the heat ".

Recipe Similarity . The final step of the online pipeline aims to retrieve from \(rel\left( SB_x\right) \) the recipe that is most similar to x . We compute the recipe-recipe similarity using BM25  (Robertson & Walker, 1994 ), a Bag-of-Word Information Retrieval model. Our main motivation for using BM25 instead of a LLM fine-tuned for text similarity such as cpt-text  (Neelakantan et al., 2022 ) is efficiency. Indeed, constraint \(C_5\) enforces our pipeline to work in real-time and because in some cases \(rel\left( SB_x\right) \) can contain more than 100, 000 recipes, we decided to use an efficient Bag-of-Word model. We evaluated the similarity computation time for 100 random recipes in the worst-case scenario (with 100, 000 comparisons) and we found that the computation time was on average 0.8 seconds ( \(\sigma = 0.1\) seconds) on a laptop with an Apple M1 processor. After computing the pair-wise similarities between x and all recipes in \(rel\left( SB_x\right) \) , we return the recipe with the highest similarity.

Experimental Design

To evaluate RELEX , we conducted a controlled user study, where we asked participants to complete three procedural writing tasks in the domain of cooking recipes using our system. In the following, we will describe the study design, participants, procedure, and the employed measures in detail.

We employed a fully randomized between-subjects design, encompassing two main factors: feedback type (adaptive vs. non-adaptive) and reflection guidance (with vs. without prompts). This resulted in four distinct treatment groups, each experiencing a specific combination of feedback and reflection instructions. To provide a basis for comparison, we also included a control group ( CG ), which received general static rules on how to write a cooking recipe, representing the traditional approach to support recipe writing without the provision of a peer example. The subjects were randomly assigned to one of the five conditions. The experiment task and questions were exactly the same for all groups; we only manipulated the adaptivity and the reflective prompts between participants. The adaptive feedback encompasses the tailored example recipe along with personalized in-text highlighting and explanations; and the reflective prompts refers to the Reflective Space where the learner is promoted to compare the recipes.

Each group used a different version of RELEX . Figure 3 shows the interface for the four treatment groups experiencing varying levels of adaptive feedback and reflection guidance. The grid has two axes: Reflective Prompts and Adaptive Feedback. Each axis has two options: With and without. Thus, each quadrant represents a distinct group differentiated by the presence or absence of adaptive feedback and reflection prompts. The interface for \(G_{R}^{A}\) including reflective prompts and adaptive feedback is displayed in the upper left quadrant. \(G_{R}^{A}\) used RELEX with all relevant functionalities including adaptive feedback (i.e., tailored example annotated with personalized in-text highlighting and explanations) and reflection prompts. The interface for \(G_{NR}^{A}\) is shown in the upper right quadrant. Accordingly, \(G_{NR}^{A}\) used RELEX without the reflection prompts. Next, as seen in the lower left quadrant, \(G_{R}^{NA}\) used RELEX without adaptive feedback, but with reflection prompts. Lastly, \(G_{NR}^{NA}\) (lower right quadrant) without reflective prompts and without adaptive feedback. Subjects in this group were displayed a pre-selected recipe from the database. Specifically, we pre-selected five complete (in terms of structure and level of detail, see “ Personalized Example Retrieval Pipeline ”), but not perfect recipes (in terms of stars, see also “ Personalized Example Retrieval Pipeline ”) from the database. We chose to not display perfect recipes in order to keep the impression of a peer recipe. Furthermore, we made sure that the five pre-selected recipes covered a range of cooking methods (e.g., dessert, hot dish, etc.). Finally, the CG did not see an example recipe; instead, an instruction manual on how to write recipes was displayed in the right pane of the tool.

Illustration of the study setup using a randomized 2 (feedback type: adaptive vs. non-adaptive) x 2 (reflection guidance: with vs. without prompts) between-subjects design resulting in four treatment groups

Participants

We recruited 200 paid participants from Prolific to conduct a controlled experiment. We chose Prolific as a platform since past research on behavioral research platforms reported the highest response quality and sample diversity for Prolific (Peer et al., 2017 ). To avoid an overlarge diversity in our sample, we recruited participants in the age range from \(18-30\) with at least an undergraduate degree as the highest completed education level. We excluded participants who did not complete the post-test or had technical problems, remaining with 187 participants for our analyses. Table  3 summarizes the demographic information per group. We did not find significant differences between the groups in terms of age ( \({\chi }^2(4) = 1.07, p = .90\) ) or gender ( \({\chi }^2(8) = 7.49, p = .48\) ) as indicated by a non-parametric Kruskal-Wallis test Footnote 10 . The median time spent on the study was 70 minutes. Participants were paid \(9\pounds \) per hour.

The experiment consisted of three main parts that were the same for all groups: (1) a pre-survey (including a pre-test), (2) three procedural writing tasks (in the domain of cooking recipes), and (3) a post-survey (including a post-test). Different from the three main tasks centered on composing cooking recipes, the pre-test and post-test were situated in a distinct domain: furniture assembly. The different domain was chosen in order to study whether the users could transfer the acquired procedural writing skills to another task.

Pre-Survey . The experiment began with a pre-survey, where we a) controlled the effectiveness of the randomization using two different constructs (see Table 4 ) and b) conducted a pre-test for procedural writing skills in the domain of furniture assembly. We started by asking each participant three questions about their previous cooking experience and documenting their recipes. Next, we captured participants’ attitudes towards technology (Agarwal & Karahanna, 2000 ). Both constructs were measured on a 7-point Likert scale (1: totally disagree to 7: totally agree, with 4 as a neutral statement). Finally, we assessed participants’ procedural writing skills in a transfer domain. Specifically, we asked participants to write the instructions to assemble an IKEA piece of furniture (a TINGBY table) based on a step-to-step diagram (illustration only, no text available). Participants were requested to spend five minutes solving the task.

Procedural writing assignment . In the procedural writing part of the experiments, we asked the participants to perform three cooking recipe writing tasks. The task was: "It’s a Sunday afternoon and your best friend calls you with a very hectic voice: they need to prepare a dish for their family who is going to visit in the evening. Your friend asks you to provide them with three different cooking recipes to choose from. Be aware that your friend has very little cooking experience and therefore you have to write the recipe as structured and understandable as possible." All groups were asked to watch an introduction video on the usage of the tool before the first recipe-writing task.

Post-Survey. The experiment ended with the post-survey. First, we conducted the post-test, where participants were asked to write instructions on how to assemble a different piece of IKEA furniture (an EKET cube) based on a step-by-step diagram (illustration only). We made sure that the difficulty of assembly was similar for both tests. As in the pre-test, participants were asked to spend five minutes on the task. Next, we measured the users’ perception using different constructs from literature (see Table 4 ). Again, all behavioral constructs were measured on a 7-point Likert scale (1: totally disagree to 7: totally agree, with 4 a neutral statement). Finally, participants answered five qualitative questions about the usage of the tool, the impact of RELEX on participants’ recipe writing, and user experience.

Measures and Analysis

To investigate the impact of our system, we studied learners’ writing performance on the task and the transfer task. Moreover, the impact on learners’ perception was assessed using a post-survey with qualitative questions. Finally, we assessed the impact on learners’ revision behavior using a keystroke analysis.

Task Performance . To assess participants’ progress in recipe writing skills, we used each participant’s first recipe (i.e., their first submission) as an initial evaluation and their last revised recipe (i.e., their last submission) as a final evaluation. Specifically, we computed two scores for each recipe: the predicted stars ( \(S_y(x)\) ) and the quality score ( \(Q_y(x)\) ). The first score, the predicted stars ( \(S_y(x)\) ), was obtained using the model fine-tuned to predict the recipes’ stars ( RELEXset-Predictor , see “ Offline Training and Annotation ”). We gave as input the recipe written by the participant and the model returned the predicted stars. The second score is a quality/completeness score based on the quality criteria (structure, clarity, specificity) implemented by the set of suggestions derived in “ Offline Training and Annotation ”. We computed the quality score \(Q_y(x)\) for a recipe x from a participant y based on \(A_{x}\) , the set of suggestions relevant to recipe x . For each suggestion \(r_i \in A_{x}\) , we computed a score \(s_{y,r_i} \in \{0,1\}\) , where 1 indicates that the suggestion was followed and 0 indicates that the suggestion was missing. We then computed the quality score as \(Q_y(x)=\sum s_{y,r_i}/|A_x|\) . The quality score, therefore, measures the ratio of followed rules for a recipe.

Transfer Performance . To evaluate the pre-and post-test tasks, we assessed the learning objectives of procedural texts. We thus adopted the subset of suggestions regarding structure, clarity, and specificity described in Table 2 . We made adjustments to \(r_8\) and \(r_9\) to better suit the context of furniture assembly. Specifically, for the specificity of materials ( \(r_8\) ), we examined the level of detail provided in describing the materials, such as explicitly naming them as wood or metal. For the specificity of steps ( \(r_9\) ), we assessed how accurately the components were referred to, including terms like screws, pegs, grooves, and knobs. Similar to measuring task performance in terms of suggestions, for each relevant suggestion \(r_i\) with, \(i \in \{1,...,9\}\) , we computed a quality score \(s_{y,r_i} \in \{0,1\}\) , where 1 indicates that the requested suggestion is followed and 0 indicates that the suggestion is missing. The overall transfer score of the task was then calculated as \(T_y(task)=\sum s_{y,r_i}/9\) .

Perception . We analyzed participants’ open responses with topic modeling. We used BERTTopic  (Grootendorst, 2022 ), a technique that incorporates the contextual information of the text by clustering embeddings generated by pre-trained transformer-based language models. We used Sentence-BERT  (Reimers & Gurevych, 2019 ) to embed the sentences in the fixed-size representation required by BERTTopic . More specifically, we used all-mpnet-base-v2 checkpoint from HuggingFace’s Transformers. We split the participants’ answers into sentences v and clustered them to obtain the topics z . The topics extracted by BERTTopic are described in terms of the most important words and their relevance. We interpreted them and assigned names to each cluster. In a next step, we computed for each sentence v the probability \(p_{v,z}\) of belonging to each cluster z . We considered that a sentence v belongs to a cluster z if \(p_{v,z} > 0.3\) to allow for sentences to be categorized into at most three topics. We then grouped the sentences by participant y to obtain the set of topics \(Z_y\) for their entire text answer. As an example, assume that the answer of a participant y consisted of three sentences \(v_1\) , \(v_2\) and \(v_3\) with assigned topics: \(v_1\) - topics A , B , \(v_2\) - topic B , and \(v_3\) - topics A , C , D . In this case, the set of topics associated with the text answer of participant y is \(Z_y = {A,B,C,D}\) .

Revision Behavior . To study users’ revision behavior, we analyzed the changes made to their recipes after receiving feedback. Based on this feedback, participants were instructed to refine their recipe. This process of analysis and improvement was not limited to a single iteration; participants could engage in multiple cycles of revision. Thus, we define a "revision" to be the set of edits (deletion, insertion, and changes) executed after receiving feedback on the recipe submission. For example, if a user requests feedback, reviews an example recipe, and subsequently makes several changes to their recipe, we consider the sequence of modifications as a single revision. If the user then proceeds to engage with the "Analyze" function once more, making additional edits to the recipe, this subsequent round of alterations is classified as a second revision. Following previous work on revision behavior and analyzing keystrokes (Mouchel et al., 2023 ; Zhu et al., 2019 ), we computed the following two features: revision time (time spent revising) and total number of revisions (number of times recipe was edited and re-submitted).

In this study, we sought to examine the effects of adaptive feedback and reflective prompts on learners’ perception (RQ1), procedural writing skills (RQ2), and revision behavior (RQ3). To achieve this, we conducted a comprehensive analysis, both quantitative and qualitative, on the data gathered from the post-survey, procedural writing assessments, and the pre- and post-test. In the following analyses we present the p -values resulting from the analysis, the effect sizes are available at: https://github.com/epfl-ml4ed/relex/tree/main/docs/effect-sizes.pdf . In a first preparatory step, we verified the randomization by checking for differences between the five groups at the beginning of the study. A Kruskal-Wallis test Footnote 11 confirmed that there were no differences in participants’ procedural writing skills as measured by their quality scores \(T_y(pre)\) (see “ Measures and Analysis ”) achieved on the pretest task ( \({\chi }^2(4) = 4.85\) , \(p = .30\) ). For the pre-survey, we obtained the construct score by averaging the items in each construct (all factor loadings were greater than 0.7) and found no significant differences either in participants’ previous experience with documenting cooking recipes ( \({\chi }^2(4) = 4.83\) , \(p = .30\) ) and attitudes towards technology ( \({\chi }^2(4) = 4.2\) , \(p =.37\) ). Lastly, we analyzed how long participants took to complete the study. On average, participants took 73 minutes. Again, we found no significant differences ( \({\chi }^2(4) = 8.15\) , \(p = .09\) ) between the average duration time per group (78 minutes for \(G_{R}^{A}\) ; 73 minutes for \(G_{NR}^{A}\) ; 64 minutes for \(G_{R}^{NA}\) ; 79 minutes for \(G_{NR}^{NA}\) ; and 70 minutes for CG .)

RQ1: Impact on Learners’ Experience

To answer our first research question, we analyzed participants’ user experience and perception. Based on the findings of Schworm and Renkl ( 2006 ), we hypothesized that the perceived learning gain, usefulness, behavioral intention, and attitude towards use would be higher in the groups with adaptive feedback: \(G_{R}^{A}\) , \(G_{NR}^{A}\) (H1-1) . In addition, in line with Venkatesh and Bala ( 2008 ), we hypothesized that the perceived ease of use would be the highest in the CG and the lowest in \(G_{R}^{A}\) given that the CG used the version with the simplest interface and functionality (H1-2) .

Quantitative Analysis . In a first analysis, we compared the post-survey constructs between groups using the Kruskal-Wallis test 11 . The results confirmed significant differences between groups concerning perceived usefulness ( \({\chi }^2(4) = 14.30\) , \(p < .01\) ) and behavioral intention ( \({\chi }^2(4) = 14.20\) , \(p < .01\) ). To further investigate the specific differences within these constructs, we performed a pairwise comparison using the Wilcoxon Rank Sum test, correcting for multiple comparisons via a Benjamini-Hochberg (BH) procedure.

Figure 4 depicts the distribution per group and construct, with statistically significant differences marked with * ( \(p < .05\) ) and ** ( \(p < .01\) ). We observe that participants from the group receiving both adaptive feedback and reflective prompts ( \(G_{R}^{A}\) ) perceived the tool as more useful than the participants from the groups without adaptive feedback ( \(G_{R}^{NA}\) and \(G_{NR}^{NA}\) ) and the control group ( CG ). Likewise, participants in \(G_{NR}^{A}\) (adaptive feedback, no reflective prompts) also reported higher perceived usefulness than participants in \(G_{R}^{NA}\) . It is worth mentioning that the only variant between these two groups was the presence of adaptive feedback. Moreover, regarding the behavioral intention, both \(G_{R}^{A}\) and \(G_{NR}^{A}\) (the groups with adaptive feedback) exhibit significantly higher scores than all other groups.

Post-survey answers comparison between control and treatment groups. Statistically significant differences between groups are indicated with * ( \(p < .05\) ) and ** ( \(p < .01\) )

In a subsequent analysis, we investigated the differences between the groups that received adaptive feedback ( \(G_{R}^{A}\) and \(G_{NR}^{A}\) ) and the ones that did not ( \(G_{R}^{NA}\) and \(G_{NR}^{NA}\) ). We found that the groups with adaptive feedback had significantly higher scores in four out of five constructs: perceived usefulness ( \({\chi }^2(1) = 11.46\) , \(p < .001\) ); attitudes toward use ( \({\chi }^2(1) = 5.2\) , \(p < .01\) ); behavioral intention ( \({\chi }^2(1) = 12.08\) , \(p < .001\) ); and perceived learning gain ( \({\chi }^2(1) = 6.07\) , \(p < 0.01\) ). Interestingly, there were no significant differences in the perceived ease of use.

Perception Analysis . In our subsequent analysis, we delved into participants’ open-text responses to gain deeper insights into the observed effects from the post-survey. Specifically, we first examined the responses to the question "What did you like?". The responses reflected a positive reception of the system’s features, including comparative viewing of recipes, in-text highlighting, ease of use, helpful suggestions, and educational insights. The most frequently mentioned aspect, noted by 16% of participants, was the opportunity to see other recipes. This feature was particularly appreciated for its comparative aspect, as highlighted by a participant from \(G_{NR}^{NA}\) : "[I liked] that I could compare my recipe with another, which makes you want to improve yours to a higher standard." The next notable aspect was in-text highlighting, valued by 11% of participants. A participant from \(G_{R}^{A}\) described this feature as "useful to quickly identify areas, and it helps you learn and observe things you can improve quite intuitively." Ease of use was also a significant point of appreciation. Participants described the system as "really intuitive, user-friendly" and "clear, easy to use and methodical ." Additionally, participants praised the quality of the suggestions offered. Comments like "I liked that it gave useful suggestions that are actually valuable to a beginner" and "It gives me tips and advice on how I can improve the wording and formatting of my recipe, so I can easily make these changes to improve the clarity and how clear my recipe is" were common. Finally, the educational insights provided by the system were highlighted. One participant mentioned, "it allowed me to gain a better perspective on how to write instructions in a clearer and more concise manner. It helped me to focus on problem areas that I subconsciously missed because it has become ingrained into my writing style. Overall, I would say that it made me more aware of my writing foibles and allowed me to thus tackle those problems and improve." Another added, "Despite reading a lot of recipes in the past, I do think that it very quickly guided me to writing more concise and easier to understand instructions. I like how quickly I learned using it, as well as how it leads you to figure out how to write good instructions rather than simply telling you a strict set of rules you must use."

Next, we examined participants’ feedback on potential improvements to the tool. Not surprisingly, \(12\%\) of participants in the control group ( CG ) proposed personalized content. One participant suggested, "I would change the recipe suggestions to be directly relevant for each written recipe. For example, after the first recipe, I added numbers to each step in the following two recipes, but still received the same feedback, so it became less useful." Similarly, \(16\%\) and \(4\%\) of the participants in \(G_{NR}^{NA}\) and \(G_{R}^{NA}\) , respectively, which were shown pre-selected recipes without using our pipeline, mentioned "adaptivity" as a potential area of improvement, suggesting to: "Limit the returned recipes to related dishes only." Interestingly, some participants in \(G_{R}^{A}\) , where participants received semantically similar examples, also expressed a desire for even more similar examples. One participant noted: "I would offer example recipes that have the same ingredients as the user’s recipe." . Another participant added "I may improve my recipe by adding ingredients that I did not previously add before to make it taste better." Furthermore, practical suggestions for future tool iterations included the ability to scan handwritten recipes, eliminating the need to retype them, and the integration of real-time tips and advice during recipe composition.

In summary, participants who received personalized examples ( \(G_{R}^{A}\) and \(G_{NR}^{A}\) ) reported significantly higher perceived usefulness, attitudes toward use, behavioral intention, and perceived learning gain compared to the other conditions, confirming (H1-1) . Interestingly, participants in the control group ( CG ), unaware of the other conditions, suggested the incorporation of adaptive feedback and content personalization, while participants in groups \(G_{R}^{NA}\) and \(G_{NR}^{NA}\) recommended showing more tailored and similar recipes. However, contrary to our expectations, there were no significant differences in the perceived ease of use between the conditions. As a result, we reject (H1-2) and conclude that the example-selection pipeline does not impose any perceivable burden or complexity on users.

RQ2: Effect on Learners’ Writing Performance

To answer the second research question, we analyzed learners’ writing performance (quantitatively) and participants’ open-text answers (qualitatively). We analyzed the users’ change in performance on the recipe task as well as on the furniture assembly task (transfer task). For the in-task performance, we hypothesized that learners who received adaptive feedback would outperform those who did not, because the highlighted elements and explanations reduce the cognitive load needed to capture the main elements (Sweller, 1994 ), enabling participants to learn faster and perform better on the task (H2-1) . In contrast, for the performance on the transfer task, we hypothesized that the participants who received reflective prompts would perform better, because of the generation effect that states that self-generated information is better retained and learned (Renkl, 2002 ) (H2-2) .

Effect on learners’ task performance . To test H2-1 , we used a repeated-measures ANOVA for the predicted stars \(S_y(x)\) and quality score \(Q_y(x)\) (see “ Measures and Analysis ”) with the conditions ( \(G_{R}^{A}\) , \(G_{NR}^{A}\) , \(G_{R}^{NA}\) , \(G_{NR}^{NA}\) and CG ) as the between-subjects and the test time (pre-score, post-score) as a within-subject factor. Subsequently, we proceeded with pairwise comparisons using the Wilcoxon Rank Sum test with BH corrections to investigate the differences between the various conditions.

In the quality score ( \(Q_y(x)\) ) analysis, we found a significant effect of test time ( \(F(1,186)=84.4, p<.0001\) ). Test time refers to the different measurements of the quality score through time, i.e., how the scores change from the first to the last recipe. Thus, a significant effect of test time means that the quality scores changed significantly over the course of the experiment. As seen in Fig.  5 (top left), the scores in general increased from the first to the last recipes. In addition, there was also a significant interaction effect ( \(F(4,186)=2.6, p<.05\) ), which indicates that the effect of time on quality scores differed depending on the experimental condition. This is also visible in Fig.  5 (top left) where some groups exhibit a steeper slope than others. This is further reinforced by a non-significant condition factor in the between-subjects analysis ( \(F(4,186)=1.65, p=.10\) ), which suggests that there were no inherent differences between the participants in the different groups. Planned pairwise comparisons confirmed the observed differences in Fig.  5 (top left). The users in \(G_{R}^{A}\) improved significantly more than the users in \(G_{R}^{NA}\) ( \(p<.05\) ). Likewise, users in \(G_{NR}^{A}\) performed significantly better than users in \(G_{R}^{NA}\) ( \(p<.05\) ) and CG ( \(p<.05\) ).

In a subsequent analysis, we investigated the differences between the groups with and without adaptive feedback (Fig. 5 top middle) as well as with and without reflective prompts (Fig. 5 top right). Planned comparisons revealed that the users with adaptive feedback improved significantly more than the users without ( \(p<.01\) ) from the first to the last recipe.

Regarding the predicted stars ( \(S_y(x)\) ) analysis, we found a significant effect of test time, with participants’ predicted stars improving significantly across recipes ( \(F(1,186)=19.2, p<.0001\) ). There was no main effect of the condition, and planned comparisons revealed no differences between the conditions.

Performance on recipe task (in terms of quality score) and transfer task. The error bars show the standard deviation

Effect on learners’ transfer performance . In a next analysis, we also used a repeated-measures ANOVA to assess performance improvements on the transfer task. Figure 5 (bottom left) illustrates the score change between participants’ pre- and post-test for the five conditions. While the CG seems to do worse than the other four conditions, we only found a significant effect of test time ( \(F(1,186)=104, p<.0001\) ). This suggests that on average, all the participants improved on the transfer task (see Fig. 5 (bottom left)). For example, \(24\%\) of the participants, who did not enumerate the steps in the pre-test, enumerated the steps in the post-test. Moreover, we reviewed the tests and noted that only two participants included a title in the pre-test, while 26 participants added it in the post-test. We also investigated the differences between the groups with and without adaptive feedback (Fig. 5 bottom middle) as well as with and without reflective prompts (Fig. 5 bottom right) and found no significant effects.

Perception Analysis . To relate the observed effects on performance to participants’ perceived performance, we again examined the survey’s open-text answers. After each recipe, participants were asked to describe the changes they made in their recipes. \(20\%\) of the participants referred to enumerating: "I numbered the steps to make the order clearer. It was a good point and will allow who is cooking to quickly find the step they need" ( \(G_{NR}^{A}\) ). Most of the consecutive popular topics referred to the recipe suggestions and explanations, for example, "specifying the size and type of pan" ( \(10\%\) ), "using more appropriate terms than add like mix, stir, beat" ( \(9\%\) ). Interestingly, despite not having direct suggestions, some participants in \(G_{R}^{NA}\) , \(G_{NR}^{NA}\) , and CG made similar changes. For example, a participant in \(G_{R}^{NA}\) mentioned: "I added a size measurement to my description of a baking pan because I realised it is helpful to have these details available for new bakers who are unsure of what sizes these things ought to be" .

In addition, as observed in the post-test, no participant in the CG mentioned adding a title and only \(3\%\) of the participants in \(G_{NR}^{A}\) mentioned it. In comparison, \(12\%\) and \(13\%\) of the participants from \(G_{R}^{NA}\) and \(G_{NR}^{NA}\) said they added a title; a participant from \(G_{R}^{NA}\) wrote: "I [originally] did not give my recipe a title. I saw that in the example recipe and realised stating the title would help the presentation" .

Additionally, to comprehend the impact of the reflective prompts, we examined how participants in groups \(G_{R}^{A}\) and \(G_{R}^{NA}\) responded regarding their utilization of these prompts. Among the participants, \(27\%\) mentioned that the prompts were useful in identifying areas of improvement, with one participant expressing, "I had to actually think about where I was going wrong and what was good about the example" . For \(12\%\) of the participants, the reflective prompts acted as a means of introspection, leading them to consider ways to enhance their own recipe writing. One participant explained, "It forced me to be introspective about my own recipe writing and thus think of ways to improve my instructions." However, a small percentage (7%) of the participants expressed a dislike for the prompts. For instance, one participant conveyed, "Not much, the reflective questions were just a part to write what I was already thinking." This observation could provide some insight into why we did not observe a significant effect of the reflective prompts on performance.

In summary, our findings support H2-1 as we observed significant differences in task performance between groups with and without adaptive feedback. However, contrary to our expectations, we did not find any significant differences in task performance between groups with and without reflective prompts, leading us to reject H2-2 . Furthermore, the results from the perception analyses indicate that participants from all groups demonstrated a good understanding of the basic elements of a procedural text.

RQ3: Effect on Learners’ Revision Behavior

In addressing our final research question, we studied how users revised their recipes after receiving feedback. We formulated two hypotheses to explore this aspect. Firstly, we hypothesized that the groups with reflection prompts would invest more time in the revision process. Participants in these groups were required to answer the reflective questions, and we anticipated that this reflective practice would lead them to approach revisions with a critical mindset, spending more time contemplating potential improvements ( H3-1 ). Additionally, we hypothesized that groups receiving adaptive feedback would continue revising over time, as the feedback provided would remain pertinent and applicable to their writing efforts ( H3-2 ).

Quantitative Analysis . Firstly, we investigated users’ average revision time (time spent revising the recipes). We compared the revision features between groups using the Kruskal-Wallis test 11 , confirming that there were significant differences between groups for time ( \({\chi }^2(4) = 12.2\) , \(p < .01\) ). Then, to investigate the differences between groups, we performed post-hoc Wilcox pairwise comparison Footnote 12 .

We found that users in group \(G_{NR}^{NA}\) spent significantly less time revising than users in \(G_{R}^{A}\) ( \(p < .05\) ) and \(G_{NR}^{A}\) ( \(p < .05\) ). However, we did not find any significant difference between the groups with and without reflecting prompts, thus rejecting H3-1 .

Next, we examined how the time spent varied between the three recipes users wrote. Figure 6 (left) illustrates the revision times of all five conditions for their first, second and third recipe. We observe that over time, users from all groups spend less time revising. It is worth noting that in the first recipe the users in \(G_{R}^{A}\) spent on average more than twice as much time (346 seconds) as the users in \(G_{R}^{NA}\) (166 seconds, \(p < .05\) ), suggesting that the adaptive features prolongated the time users revised the recipes.

Revision behavior: time spent revising, number of revisions, and the percentage of declared no changes

Furthermore, when analyzing the number of revisions, we also found significant differences in the overall number of revisions per group ( \({\chi }^2(4) = 23.6\) , \(p < .001\) ). In particular, group \(G_{NR}^{A}\) revised their recipes more than the rest of the groups ( \(G_{R}^{A}\) , \(p < .05\) ; \(G_{R}^{NA}\) , \(p < .001\) ; \(G_{NR}^{NA}\) , \(p < .001\) ; CG , \(p < .01\) ).

Moreover, we examined the number of revisions per recipe and found that there was also a general declining trend in the average number of revisions (see Fig. 6 (middle)). Analogously to the general results, in the first recipe, the users in \(G_{NR}^{A}\) revised their recipe significantly more than the groups with no adaptive feedback ( \(p < .01\) for \(G_{R}^{NA}\) and \(G_{NR}^{NA}\) ). Likewise, in the second recipe, \(G_{NR}^{A}\) had significantly more revisions than \(G_{R}^{NA}\) ( \(p = .005\) ), \(G_{NR}^{NA}\) ( \(p = .006\) ) and CG ( \(p = .02\) ). Despite the fact that users in \(G_{NR}^{A}\) also reduced their revision count throughout all three recipes, they consistently maintained a higher average number of revisions compared to the other groups. This finding supports hypothesis H3-2 , indicating that certain users who received adaptive feedback still perceived it as interesting or valuable enough to ask for it again. Nonetheless, it is notable that for the first recipe, users in \(G_{NR}^{A}\) revised more than users in \(G_{R}^{A}\) , despite both having adaptive feedback. It might be possible that the reflective prompts increased the cognitive load for \(G_{R}^{A}\) , leading to less revisions.

Perception Analysis . As mentioned earlier, after each submission, participants were asked to describe the changes they made to improve their recipes. Figure 6 (right) shows the percentage of participants reporting not making any changes for their first, second, and third recipes. We observe that for all groups, a large majority of users reported changes, with the percentage of participants not improving their recipe, increased from the first to the last recipe. Not surprisingly, group CG had the steepest increase: \(29\%\) of participants in this group reported making no changes to their last recipe. One participant in this group mentioned that The Analyze button just outputs the same suggestions every time, so I knew already what it wanted, and I didn’t need to make any changes . This suggests that the feedback became redundant as it was static and there were no changes. In contrast, \(88\%\) ( \(G_{R}^{A}\) ) and \(86\%\) ( \(G_{NR}^{A}\) ) of the participants in the adaptive feedback groups continued to report changes they made to the recipe. A big portion of the changes reported by the users (83%), came from (or were very similar to) the suggestion given by the system. Interestingly, in the first recipe, most changes were related to the structure of the recipe, for example: "I added the ingredients list and made it step by instructions. I made these steps to make it easier to follow." Whereas in the second and third recipes, most comments referred to the specificity of the instructions and the steps, for example, one participant of \(G_{NR}^{A}\) mentioned: "I described exactly when to move onto a next step and what to look out for in a mixture in order to proceed" .

In summary, we reject H3-1 as we did not see the groups with reflection prompts spending more time revising. Moreover, our quantitative and qualitative analyses support H3-2 indicating that groups with adaptive feedback perceive the example recipe and annotations as relevant, while suggestions for the other groups started to feel redundant.

Discussion and Conclusion

In this paper, we presented RELEX , an adaptive learning system for enhancing procedural writing skills. RELEX features a real-time retrieval pipeline, enabling personalized example-based learning at scale. Our multi-step pipeline selects higher quality and semantically relevant examples for learners based on their input and provides suggestions on how to improve their writing. We evaluated RELEX with 200 users to analyze the effects of personalized examples and reflective prompts on users’ writing performance, perceived experience, and revision behavior.

Impact on learners’ experience (RQ1) . Our results show that providing adaptive feedback on procedural writing skills has a positive impact on the user experience (RQ1). As we hypothesized ( H1-1: Adaptive feedback will lead to heightened perceptions of learning gain, usefulness, behavioral intention, and more positive attitudes towards usage among learners ), learners who received personalized recipes and adaptive feedback ( \(G_{R}^{A}\) and \(G_{NR}^{A}\) ) judged the perceived learning gain, the perceived usefulness, the behavioral intention for continuous use, and the attitude towards use significantly better than those who did not receive adaptive feedback ( \(G_{R}^{NA}\) and \(G_{NR}^{NA}\) ). These results are coherent with previous work (Wambsganss et al., 2020 ), where the group with adaptive feedback had a significantly higher intention to use. Moreover, our analysis of open answers exemplifies the positive reactions participants had towards seeing another recipe, in-text highlighted elements, and adaptive suggestions. A positive perception plays an important role in the long-term success of learning tools and their potential to foster learning (Kirkpatrick, 1994 ).

Against our expectations and different from Fan et al. ( 2017 ), we did not find any significant differences between the groups regarding the perceived ease of use ( H1-2: The ease of use will be perceived as most favorable in the groups with simpler interfaces ). We originally hypothesized that the users would find the complete interface (including the personalized example, adaptive explanations, in-text highlighting, and reflective prompts) hard to understand. Venkatesh and Bala ( 2008 ) define perceived ease of use as the degree to which a person believes that using the tool with be free of effort. Thus, we expected the extra features like reflection and suggestions to represent an effort for the users. Nevertheless, when analyzing the qualitative comments, the third highest-ranked topic was the "intuitiveness" of the tool. This suggests that the design iterations with users contributed to an intuitive design, where the special features and elements do not hinder the ease of use.

Impact on learners’ writing performance (RQ2) . Moreover, we investigated the effects of the design elements (personalized example, adaptive feedback and prompts) on performance (RQ2). Our results confirm our hypothesis ( H2-1: Adaptive feedback will improve in-task writing performance ), showing that participants in the adaptive feedback groups improved their recipe quality and completeness significantly more than the participants in the non-adaptive groups. The perception analysis suggests that the in-text highlighted elements helped identify the areas of opportunity quickly. Previous work (van Gog et al., 2008 ) found that extra information and explanations were beneficial in terms of learning gains at first, but hindered performance later on as the information quickly became redundant. In our study, we overcome that challenge in \(G_{R}^{A}\) and \(G_{NR}^{A}\) by only showing explanations that are relevant based on the user’s recipe. This adaptivity could also explain the observed performance differences given that \(G_{R}^{NA}\) - CG received redundant explanations regardless of the user’s input. This is in line with the perception analysis, where the participants in the CG mentioned that the suggestions became less useful when they were redundant.

We also studied whether participants in the groups with reflective prompts were able to generalize better when asked to transfer the skills to another domain ( H2-2: Reflective prompts will improve the writing performance in a transfer task ). Our results reject our hypothesis. We hypothesize that the duration of the user study was too short (only three recipes) to unfold the self-explanation effect (Wong et al., 2002 ). Alternatively, as noted by one of the participants, it is possible that even without writing, the participants were already explaining the example to themselves.

Surprisingly, all groups improved on the transfer task (furniture assembly). We observed that, on average, participants improved their text \(15\%\) in terms of quality (structure and specificity). This suggests that participants were able to grasp the principles of the learning domain (procedural writing) and apply them to a different exemplifying domain (furniture assembly). Furthermore, our results from H2-1 and the perception analysis indicate that participants also learned elements specific to the cooking domain (e.g., specifying the heat intensity). In H2-1 we observed significant differences when measuring the improvements from both content levels. We therefore hypothesize that the five approaches (experimental conditions) are similarly effective in teaching general procedural writing skills (from the learning domain). Yet, the conditions incorporating adaptive feedback also enhance learners’ understanding in a specialized area within procedural writing: cooking recipe writing (exemplifying domain). On average, participants improved the structure and organization of their procedural text by \(15\%\) , including enumerating the steps, listing the materials, having separate sections for materials and steps, and adding appropriate sub-headings. According to the double-content description provided by Renkl et al. ( 2009 ), these elements belong to the content level of the learning domain of procedural writing (i.e., how to structure a procedural text in general), i.e. participants were able to grasp the fundamental structural elements of procedural writing by practicing only in the example domain.

Impact on learners’ revision behavior (RQ3) . In the last analysis, against our expectations, we did not find that the use of reflective questions led to extended periods of revision. On the contrary, we found that users who received adaptive feedback spent more time revising ( H3-1: Reflective prompts will increase the duration of revision times ) than the users without adaptive feedback. Moreover, we observed that in general the time spent revising, as well as the number of revisions, decreased from the first to the last recipe. It is indeed interesting that despite the users spending less time revising, the recipes are of higher quality (as seen in RQ2 ). As the perception analysis revealed, the users made fewer changes because they had already incorporated some of the feedback. Zhu et al. ( 2020 ) also observed a decline in the revision time with multiple tasks and hypothesized that the users became more familiar with the content and the feedback resulting in less time reading feedback and making changes.

As expected, users with adaptive feedback continued to revise more in their second and third recipes ( H3-2: Adaptive feedback will result in an increased number of revisions ). The percentage of users in groups with adaptive feedback that report making no changes in the last recipe is lower than in the other groups. van Gog and Rummel ( 2010 ) observed instructional explanations becoming redundant and irrelevant over time; it seems that providing personalized examples and annotations indeed helps reducing this effect. These results complement the results from RQ1 , it is possible that the users perceived the tool as more useful if they engaged more with the feedback and spent more time making changes.

Literature Contributions . Our study contributes to and expands prior research in two main literature streams.

First, we contribute to the literature stream of artificial intelligence (AI) for example-based learning in heuristic domains. Most prior research (van Gog et al., 2008 ; van Gog & Rummel, 2010 ; Renkl et al., 2009 ; Renkl, 2002 ) on example-based learning uses static examples: both the examples and explanations are created by experts and all the learners see the exact same content, independent of their input. In contrast to past literature, RELEX provides examples tailored to the needs of the learner in terms of topic (i.e. similar content) and skill level. Instead of providing a perfect expert example, we provide a peer example of better quality, but still attainable. Furthermore, we also personalize the instructional explanations based on the input text of the learner. Additionally, we enhance the adaptive feedback by incorporating reflective prompts, leveraging the documented benefits found in the existing literature (Schworm & Renkl, 2007 ; Wong et al., 2002 ; Chi et al., 1989 ; Roelle et al., 2012 ).

Second, we contribute to the literature around SRL in AI systems. By including prompts for self-evaluation within the design of RELEX , we shed light on the combination of reflective prompts and personalized content and their effect on learning experiences and learning outcomes. Despite the qualitative comments on the helpfulness of the prompts and the positive effects from previous work (Roelle et al., 2012 ; Schworm & Renkl, 2007 ; van Gog & Rummel, 2010 ), we did not see a significant effect on our quantitative outcome variables for perception or performance. This opens new lines of future research to investigate how to best integrate reflective prompts into adaptive systems.

RELEX contrasts with previous approaches to instruct procedural writing skills by focusing on personalization and adaptivity. In comparison to previous works (Traga Philippakos, 2019 ; Sato & Matsushima, 2006 ; Alviana, 2019 ) where the instructional materials are static, meaning that all students received the same examples, in RELEX the example is chosen to cater to individual learning needs. Moreover, in comparison to instructional group approaches (Traga Philippakos, 2019 ), in RELEX each student can learn at their own pace and different from (Sato & Matsushima, 2006 ), it does not require external readers to give feedback. Furthermore, in contrast to other approaches of example-based learning (Sweller, 1994 ; van Gog et al., 2008 ; Renkl et al., 2009 ; Renkl, 2002 ), in our work, not only do we provide a personalized example, but we also offset the common disadvantage of instructional explanations being redundant or too complex. By annotating the examples with instructional explanations adapted to the learner’s prior text, we ensure their relevance.

Limitations and Future Work . One of the big challenges of enriching examples in example-based learning is the relevance of the explanations (Renkl, 2002 ). Despite the participants’ positive perception of the suggestions, they were extracted from "The Recipe Writer’s Handbook, Revised and Expanded" (Ostmann & Baker, 2001 ) and inevitably include the authors’ bias. For example, there are more suggestions for ingredients used in Western cuisine. The implication of this is that at scale, learners who write recipes from Western cuisine could benefit more from relevant suggestions. Future lines of work should investigate these biases and how to mitigate them.

Another limitation emerging from the database is that the prediction model was trained on user ratings that can be subjective. In addition, the ratings were given for a recipe as a whole, combining writing quality and taste. We examined the comments associated with the ratings and found that high-rated recipes (five stars) often had comments appreciating the clarity of instructions, as exemplified by remarks like " I really appreciate the instructions about using the spoon when cutting the potatoes. This is a well-written recipe." ; and "This was easy enough to prepare on a worknight and assembly was so easy when following the well-written directions" . Conversely, recipes with low ratings were often criticized for their lack of clarity and order, as indicated by comments such as "This recipe is written in a way that is impossible to attempt to follow or understand. It is a disaster." ; and "Very frustrated with the directions. They are not orderly whatsoever." . This suggests that even if a recipe is tasty, unclear writing can hinder its reproducibility, leading to low ratings. However, we acknowledge that a recipe with excellent writing but an unfamiliar or unappealing taste might also receive low ratings. In future studies, it would be beneficial to separate the variables taste and writing quality to more accurately assess their individual impacts on user ratings.

This complexity extends to the predictive task, where RELEXset-Predictor attempts to account for both taste and writing quality, leading to only minor improvement over a static baseline. We have therefore made our code and models publicly available Footnote 13 , encouraging future research to enhance predictive accuracy, for example through the integration of new SOTA models. The design of the subsequent stages of the pipeline attempts to mitigate the limitations of RELEXset-Predictor . Overall, participants perceived the adaptive recipes as useful and edited the recipes accordingly. However more rigorous, quantitative assessments are needed to investigate the influence of the model performance and the chosen quality range on user perception. Furthermore, RELEX offers a promising approach for learners to improve their recipe writing skills by integrating both the learning domain (procedural writing) and the exemplifying domain (cooking). Despite its effectiveness, its scope is limited to these specific areas. One main takeaway for the research community is the demonstrated importance of adaptivity and personalization in example-based learning, particularly in enhancing user engagement and performance outcomes. In future work, the example-selection pipeline can be adapted to cater to other learning domains. For instance, journal writing (Roelle et al., 2012 ), high school instruction (Hilbert et al., 2008 ), or argumentative writing (Schworm & Renkl, 2007 ). Transferring RELEX to a different exemplifying or learning domain requires two main ingredients: 1) multiple examples with associated evaluations, ratings, or grades, and 2) domain-specific suggestions regarding example annotation. The selection pipeline (see “ Personalized Example Retrieval Pipeline ”) can be used to fine-tune an NLP model to predict the evaluations of the examples. Then, the model name and domain-specific suggestions can be added to the code base of RELEX to run the application. By extending the tool’s capabilities to various educational contexts, we anticipate a broader impact and potential benefits for learners across different domains Footnote 14 .

In the future, we envision expanding the scope and applicability of our findings by conducting replication studies in real-world settings, such as classrooms with chef apprentices. This approach would help address the ecological validity of the results and provide insights into the effectiveness of RELEX in practical educational contexts. Additionally, we plan to explore the long-term effects of RELEX by conducting a longitudinal study, assessing how repeated usage of the tool impacts learners’ procedural writing skills over an extended period.

The demo version of RELEX is available at https://go.epfl.ch/relex

The detailed interview questions can be found on https://github.com/epfl-ml4ed/relex/blob/main/docs/user-interviews.pdf

RELEXset can be downloaded from https://github.com/epfl-ml4ed/relex/readme.md

RELEXset-MLM is available at https://huggingface.co/paola-md/RELEXset-MLM and RELEXset-Predictor is available at https://huggingface.co/paola-md/RELEXset-Predictor/

The architecture configuration is available at https://huggingface.co/paola-md/RELEXset-Predictor/blob/main/config.json

After a significant Shapiro-Wilk test on the three sets ( \(p=0\) )

Visual validation available at: https://github.com/epfl-ml4ed/relex/blob/main/docs/split-verification.ipynb

Complete list of suggestions and classification rules available at https://github.com/epfl-ml4ed/relex/docs/recipe-suggestions-rules.pdf .

https://norvig.com/mayzner.html

We checked for normality using a Shapiro-Wilk test and verified equal variances using Levene’s test and found that for both age and gender, the assumptions of ANOVA were not satisfied.

We checked for normality using a Shapiro-Wilk test and verified equal variances using Levene’s test and found the assumptions of ANOVA were not satisfied.

correcting for multiple comparisons via BH procedure.

For those interested in replicating or building upon our work, we have made the implementation code and instructions for domain transfer available at https://github.com/epfl-ml4ed/relex/readme.md

For those interested in replicating or building upon our work, we have made the implementation code and instructions for domain transfer available at https://github.com/epfl-ml4ed/relex/readme.md .

Adoniou, M. (2013). Drawing to support writing development in english language learners. Language and Education , 27 (3), 261–277. Retrieved from https://doi.org/10.1080/09500782.2012.704047

Afrin, T., Kashefi, O., Olshefski, C., Litman, D., Hwa, R., & Godley, A. (2021). Effective interfaces for student-driven revision sessions for argumentative writing. Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems (pp. 1–13). ACM. Retrieved from https://doi.org/10.1145/3411764.3445683

Agarwal, R., & Karahanna, E. (2000). Time flies when you’re having fun: cognitive absorption and beliefs about information technology usage. MIS Quarterly , 24 (4), 665–694. Retrieved 2022-09-13 from http://www.jstor.org/stable/3250951

Ahmed, U.Z., Srivastava, N., Sindhgatta, R., & Karkare, A. (2020). Characterizing the pedagogical benefits of adaptive feedback for compilation errors by novice programmers. In: Proceedings of the ACM/IEEE 42nd International Conference on Software Engineering: Software Engineering Education and Training (pp. 139–150). ACM. Retrieved from https://doi.org/10.1145/3377814.3381703

Alamri, H., Lowell, V., Watson, W., & Watson, S.L. (2020). Using personalized learning as an instructional approach to motivate learners in online higher education: Learner self-determination and intrinsic motivation. Journal of Research on Technology in Education , 52 (3), 322–352. Retrieved from https://doi.org/10.1080/15391523.2020.1728449

Alviana, V. (2019). The effect of recipe demonstration technique on students’ writing competence in procedural text. Journal of Languages and Language Teaching, 7 (2), 128–131.

Article   Google Scholar  

Ambarwati, S., & Listyani, L. (2021). Procedural essay writing: Students’ problems and strategies. LLT Journal: A Journal on Language and Language Teaching, 24 (2), 364–379.

Bassen, J., Balaji, B., Schaarschmidt, M., Thille, C., Painter, J., Zimmaro, D. & Mitchell, J.C. (2020). Reinforcement learning for the adaptive scheduling of educational activities. In: CHI ’20: CHI Conference on Human Factors in Computing Systems (pp. 1–12). ACM. Retrieved from https://doi.org/10.1145/3313831.3376518

Bień, M., Gilski, M., Maciejewska, M., Taisner, W., Wisniewski, D., & Lawrynowicz, A. (2020). RecipeNLG: A cooking recipes dataset for semi-structured text generation. In: Proceedings of the 13th International Conference on Natural Language Generation (pp. 22–28). ACL. Retrieved from https://aclanthology.org/2020.inlg-1.4

Bimba, A.T., Idris, N., Al-Hunaiyyan, A., Mahmud, R.B., & Shuib, N.L.B.M. (2017). Adaptive feedback in computer-based learning environments: a review. Adaptive Behavior , 25 (5), 217–234. Retrieved from https://doi.org/10.1177/1059712317727590

Brown, T.B., & et al. (2020). Language models are few-shot learners. Advances in Neural Information Processing Systems 33 . Retrieved from https://proceedings.neurips.cc/paper/2020/hash/1457c0d6bfcb4967418bfb8ac142f64a-Abstract.html

Chi, M.T., Bassok, M., Lewis, M.W., Reimann, P., & Glaser, R. (1989). Self-explanations: How students study and use examples in learning to solve problems. Cognitive Science , 13 (2), 145–182. Retrieved from https://doi.org/10.1016/0364-0213(89)90002-5

Compeau, D.R., & Higgins, C.A. (1995). Computer self-efficacy: Development of a measure and initial test. MIS Quarterly , 19 (2), 189–211. Retrieved from http://www.jstor.org/stable/249688

Cooper, A., Reimann, R., & Cronin, D. (2007). About face 3: the essentials of interaction design (3rd edition). Wiley Pub.

Davis, F.D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly , 13 (3), 319–340. Retrieved 2022-09-13 from http://www.jstor.org/stable/249008

Devlin, J., Chang, M., Lee, K., & Toutanova, K. (2019). BERT: pre-training of deep bidirectional transformers for language understanding. In: Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (pp. 4171–4186). Association for Computational Linguistics. Retrieved from https://doi.org/10.18653/v1/n19-1423

Doroudi, S., Kamar, E., Brunskill, E., & Horvitz, E. (2016). Toward a learning science for complex crowdsourcing tasks. In: Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems (pp. 2623–2634). ACM. Retrieved from https://doi.org/10.1145/2858036.2858268

Fan, X., Luo, W., Menekse, M., Litman, D., & Wang, J. (2017). Scaling reflection prompts in large classrooms via mobile interfaces and natural language processing. Proceedings of the 22nd International Conference on Intelligent User Interfaces (pp. 363–374). ACM. Retrieved from https://doi.org/10.1145/3025171.3025204

Grootendorst, M. (2022). BERTopic: Neural topic modeling with a class-based TF-IDF procedure . Retrieved from arXiv:2203.05794

Gururangan, S., Marasovic, A., Swayamdipta, S., Lo, K., Beltagy, I., Downey, D., & Smith, N.A. (2020). Don’t stop pretraining: Adapt language models to domains and tasks. Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics (pp. 8342–8360). ACL. Retrieved from https://doi.org/10.18653/v1/2020.acl-main.740

Hilbert, T.S., Renkl, A., Kessler, S., & Reiss, K. (2008). Learning to prove in geometry: Learning from heuristic examples and how it can be supported. Learning and Instruction , 18 (1), 54–65. Retrieved from https://doi.org/10.1016/j.learninstruc.2006.10.008

Hosseini, R., & Brusilovsky, P. (2017). A study of concept-based similarity approaches for recommending program examples. New Review of Hypermedia and Multimedia , 23 (3), 161–188. Retrieved from https://doi.org/10.1080/13614568.2017.1356878

Hu, G., Ahmed, M., & L’Abbé, M.R. (2022). Natural language processing and machine learning approaches for food categorization and nutrition quality prediction compared to traditional methods. The American Journal of Clinical Nutrition , 553–563. Retrieved from https://doi.org/10.1016/j.ajcnut.2022.11.022

Jin, R., & Si, L. (2004). A study of methods for normalizing user ratings in collaborative filtering. Proceedings of the 27th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (pp. 568–569). ACM. Retrieved from https://doi.org/10.1145/1008992.1009124

Kingma, D.P., & Ba, J. (2015). Adam: A method for stochastic optimization. 3rd International Conference on Learning Representations . Retrieved from arXiv:1412.6980

Kirkpatrick, D. L. (1994). Evaluating training programs: The four levels . San Francisco: Berrett-Koehler Publishers.

Google Scholar  

Landis, J.R., & Koch, G.G. (1977). An application of hierarchical kappa-type statistics in the assessment of majority agreement among multiple observers. Biometrics , 33 (2), 363–374. Retrieved from http://www.jstor.org/stable/2529786

Liu, Y., Ott, M., Goyal, N., Du, J., Joshi, M., Chen, D. & Stoyanov, V. (2019). RoBERTa: A Robustly Optimized BERT Pretraining Approach . Retrieved from arXiv:1907.11692

Majumder, B.P., Li, S., Ni, J., & McAuley, J. (2019). Generating personalized recipes from historical user preferences. In: Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (pp. 5976–5982). ACL. Retrieved from https://doi.org/10.18653/v1/D19-1613

Max, L., Alex, S., & Dmytro, L. (2022). Grammarly . Retrieved from https://app.grammarly.com/

Mayfield, E., & Black, A.W. (2020). Should you fine-tune bert for automated essay scoring? In: Proceedings of the Fifteenth Workshop on Innovative Use of NLP for Building Educational Applications (pp. 151–162). Association for Computational Linguistics. Retrieved from https://doi.org/10.18653/v1/2020.bea-1.15

Mejia-Domenzain, P., Marras, M., Giang, C., Cattaneo, A., & Käser, T. (2022). Evolutionary clustering of apprentices’ self- regulated learning behavior in learning journals. IEEE Transactions on Learning Technologies , 1–14. Retrieved from https://doi.org/10.1109/TLT.2022.3195881

Mouchel, L., Wambsganss, T., Mejia-Domenzain, P., & Käser, T. (2023). Understanding revision behavior in adaptive writing support systems for education. International Conference on Educational Data Mining , 445–452. Retrieved from arXiv:2306.10304

Nah, F.F. (2003). A study on tolerable waiting time: How long are web users willing to wait? 9th Americas Conference on Information Systems (p. 285). AIS. Retrieved from http://aisel.aisnet.org/amcis2003/285

Neelakantan, A., Xu, T., Puri, R., Radford, A., Han, J.M., Tworek, J. & et al. (2022). Text and code embeddings by contrastive pre-training. Retrieved from arXiv:2201.10005

Nückles, M., Hübner, S., & Renkl, A. (2009). Enhancing self-regulated learning by writing learning protocols. Learning and Instruction , 19 (3), 259–271. Retrieved from https://doi.org/10.1016/j.learninstruc.2008.05.002

Ostmann, B.G.O., & Baker, J.L. (2001). The recipe writer’s handbook, revised and expanded . Harvest.

Paassen, B., Hammer, B., Price, T.W., Barnes, T., Gross, S., & Pinkwart, N. (2018). The Continuous Hint Factory - Providing Hints in Vast and Sparsely Populated Edit Distance Spaces. Journal of Educational Data Mining , 10 (1), 1–35. Retrieved from arXiv:1708.06564

Peer, E., Brandimarte, L., Samat, S., & Acquisti, A. (2017). Beyond the turk: Alternative platforms for crowdsourcing behavioral research. Journal of Experimental Social Psychology, 70 , 153–163.

Pelánek, R. (2020). Measuring similarity of educational items: An overview. IEEE Transactions on Learning Technologies , 13 (2), 354–366. Retrieved from https://doi.org/10.1109/TLT.2019.2896086

Premlatha, K.R., & Geetha, T.V. (2015). Learning content design and learner adaptation for adaptive e-learning environment: a survey. Artificial Intelligence Review , 44 (4), 443–465. Retrieved from https://doi.org/10.1007/s10462-015-9432-z

Reimers, N., & Gurevych, I. (2019). Sentence-bert: Sentence embeddings using siamese bert-networks. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (pp. 3980–3990). ACL. Retrieved from https://doi.org/10.18653/v1/D19-1410

Renkl, A. (2002). Worked-out examples: instructional explanations support learning by self-explanations. Learning and Instruction , 12 (5), 529–556. Retrieved from https://doi.org/10.1016/S0959-4752(01)00030-5

Renkl, A., Hilbert, T., & Schworm, S. (2009). Example-based learning in heuristic domains: A cognitive load theory account. Educational Psychology Review , 21 (1), 67–78. Retrieved from https://doi.org/10.1007/s10648-008-9093-4

Ringenberg, M.A., & VanLehn, K. (2006). Scaffolding problem solving with annotated, worked-out examples to promote deep learning. Intelligent Tutoring Systems (pp. 625–634). Springer Berlin Heidelberg.

Robertson, S.E., & Walker, S. (1994). Some simple effective approximations to the 2-poisson model for probabilistic weighted retrieval. Proceedings of the 17th Annual International Conference on Research and Development in Information Retrieval. (pp. 232–241). ACM. Retrieved from https://doi.org/10.1007/978-1-4471-2099-5_24

Roelle, J., Krüger, S., Jansen, C., & Berthold, K. (2012). The use of solved example problems for fostering strategies of self-regulated learning in journal writing. Education Research International. (2012). 751625. Retrieved from https://doi.org/10.1155/2012/751625

Rogers, T., & Feller, A. (2016). Discouraged by peer excellence: Exposure to exemplary peer performance causes quitting. Psychological Science , 27 (3), 365–374. Retrieved from https://doi.org/10.1177/0956797615623770

Sanh, V., Debut, L., Chaumond, J., & Wolf, T. (2019). DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter . Retrieved from arXiv:1910.01108

Sato, K., & Matsushima, K. (2006). Effects of audience awareness on procedural text writing. Psychological Reports , 99 (1), 51–73. Retrieved from https://doi.org/10.2466/pr0.99.1.51-73

Schwonke, R., Renkl, A., Krieg, C., Wittwer, J., Aleven, V., & Salden, R. (2009). The worked-example effect: Not an artefact of lousy control conditions. Computers in Human Behavior , 25 (2), 258–266. Retrieved from https://doi.org/10.1016/j.chb.2008.12.011

Schworm, S., & Renkl, A. (2006). Computer-supported example-based learning: When instructional explanations reduce self-explanations. Computers & Education , 46 (4), 426–445. Retrieved from https://doi.org/10.1016/j.compedu.2004.08.011

Schworm, S., & Renkl, A. (2007). Learning argumentation skills through the use of prompts for self-explaining examples. Journal of Educational Psychology , 99 (2), 285–296. Retrieved from https://doi.org/10.1037/0022-0663.99.2.285

Sellam, T., Das, D., & Parikh, A.P. (2020). BLEURT: learning robust metrics for text generation. Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics (pp. 7881–7892). ACL. Retrieved from https://doi.org/10.18653/v1/2020.acl-main.704

Slade, C., & Downer, T. (2020). Students’ conceptual understanding and attitudes towards technology and user experience before and after use of an eportfolio. Journal of Computing in Higher Education , 32 (3), 529–552. Retrieved from https://doi.org/10.1007/s12528-019-09245-8

Sun, C., Qiu, X., & Xu, Y. (2019). How to fine-tune BERT for text classification? Chinese Computational Linguistics - 18th China National Conference (Vol. 11856, pp. 194–206). Springer. Retrieved from https://doi.org/10.1007/978-3-030-32381-3_16

Sweller, J. (1994). Cognitive load theory, learning difficulty, and instructional design. Learning and Instruction , 4 (4), 295–312. Retrieved from https://doi.org/10.1016/0959-4752(94)90003-5

Traga Philippakos, Z.A. (2019). Effects of strategy instruction with an emphasis on oral language and dramatization on the quality of first graders’ procedural writing. Reading & Writing Quarterly , 35 (5), 409–426. Retrieved from https://doi.org/10.1080/10573569.2018.1547233

van Gog, T., & Rummel, N. (2010). Example-based learning: Integrating cognitive and social-cognitive research perspectives. Educational Psychology Review , 22 (2), 155–174. Retrieved from https://doi.org/10.1007/s10648-010-9134-7

van Gog, T., Paas, F., & van Merriënboer, J.J. (2008). Effects of studying sequences of process-oriented and product-oriented worked examples on troubleshooting transfer efficiency. Learning and Instruction , 18 (3), 211–222. Retrieved from https://doi.org/10.1016/j.learninstruc.2007.03.003

Venkatesh, V., & Bala, H. (2008). Technology acceptance model 3 and a research agenda on interventions. Decision Sciences , 39 (2), 273–315. Retrieved from https://doi.org/10.1111/j.1540-5915.2008.00192.x

Wang, W., Arya, D.M., Novielli, N., Cheng, J., & Guo, J.L.C. (2020). Argulens: Anatomy of community opinions on usability issues using argumentation models. Conference on Human Factors in Computing Systems (pp. 1–14). ACM. Retrieved from https://doi.org/10.1145/3313831.3376218

Wambsganss, T., Niklaus, C., Cetto, M., Söllner, M., Handschuh, S., Leimeister, J.M. (2020). AL: an adaptive learning support system for argumentation skills. In Proceedings of the 2020 CHI conference on human factors in computing systems (pp. 1–14)

Wieringa, D.R., & Farkas, D.K. (1991). Procedure writing across domains: nuclear power plant procedures and computer documentation. Proceedings of the 9th Annual International Conference on Systems Documentation (pp. 49–58).

Wilson, J., Olinghouse, N.G., & Andrada, G.N. (2014). Does automated feedback improve writing quality? Learning Disabilities: A Contemporary Journal , 12 (1), 93–118. Retrieved from https://eric.ed.gov/?id=EJ1039856

Wolf, T., Debut, L., Sanh, V., Chaumond, J., Delangue, C., Moi, A., & Brew, J. (2019). Huggingface’s transformers: State-of-the-art natural language processing. Retrieved from arXiv:1910.03771

Wong, R.M., Lawson, M.J., & Keeves, J. (2002). The effects of self-explanation training on students’ problem solving in high-school mathematics. Learning and Instruction , 12 (2), 233–262. Retrieved from https://doi.org/10.1016/S0959-4752(01)00027-5

Zhu, M., Liu, O.L., & Lee, H.-S. (2020). The effect of automated feedback on revision behavior and learning gains in formative assessment of scientific argument writing. Computers & Education , 143 , 103668. Retrieved from https://doi.org/10.1016/j.compedu.2019.103668

Zhu, M., Zhang, M., & Deane, P. (2019). Analysis of Keystroke Sequences in Writing Logs. ETS Research Report Series , 2019 (1), 1–16. Retrieved from https://doi.org/10.1002/ets2.12247

Zlabinger, M., Sabou, M., Hofstätter, S., Sertkan, M., & Hanbury, A. (2020). DEXA: supporting non-expert annotators with dynamic examples from experts. Proceedings of the 43rd International conference on research and development in Information Retrieval (pp. 2109–2112). ACM. Retrieved from https://doi.org/10.1145/3397271.3401334

Download references

Open access funding provided by EPFL Lausanne. This project was substantially co-financed by the Swiss State Secretariat for Education, Research and Innovation (SERI).

Author information

Authors and affiliations.

School of Computer and Communication Sciences, EPFL, Lausanne, Switzerland

Paola Mejia-Domenzain, Jibril Frej, Seyed Parsa Neshaei, Luca Mouchel, Tanya Nazaretsky, Thiemo Wambsganss, Antoine Bosselut & Tanja Käser

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Paola Mejia-Domenzain .

Ethics declarations

Competing interests.

The authors have no relevant financial or non-financial interests to disclose.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Mejia-Domenzain, P., Frej, J., Neshaei, S.P. et al. Enhancing Procedural Writing Through Personalized Example Retrieval: A Case Study on Cooking Recipes. Int J Artif Intell Educ (2024). https://doi.org/10.1007/s40593-024-00405-1

Download citation

Accepted : 10 April 2024

Published : 22 April 2024

DOI : https://doi.org/10.1007/s40593-024-00405-1

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Example-based learning
• Procedural writing
• Large language models
• Text quality evaluation
• Find a journal
• Publish with us
• Track your research

HYPOTHESIS AND THEORY article

This article is part of the research topic.

Using Case Study and Narrative Pedagogy to Guide Students Through the Process of Science

Molecular Storytelling: A Conceptual Framework for Teaching and Learning with Molecular Case Studies Provisionally Accepted

• 1 School of Interdisciplinary Arts and Sciences, University of Washington Bothell, United States
• 2 Institute for Quantitative Biomedicine, Rutgers, The State University of New Jersey, United States
• 3 Research Collaboratory for Structural Bioinformatics Protein Data Bank, Rutgers, The State University of New Jersey,, United States

The final, formatted version of the article will be published soon.

Molecular case studies (MCSs) provide educational opportunities to explore biomolecular structure and function using data from public bioinformatics resources. The conceptual basis for the design of MCSs has yet to be fully discussed in the literature, so we present molecular storytelling as a conceptual framework for teaching with case studies. Whether the case study aims to understand the biology of a specific disease and design its treatments or track the evolution of a biosynthetic pathway, vast amounts of structural and functional data, freely available in public bioinformatics resources, can facilitate rich explorations in atomic detail. To help biology and chemistry educators use these resources for instruction, a community of scholars collaborated to create the Molecular CaseNet. This community uses storytelling to explore biomolecular structure and function while teaching biology and chemistry. In this article, we define the structure of an MCS and present an example. Then, we articulate the evolution of a conceptual framework for developing and using MCSs. Finally, we related our framework to the development of technological, pedagogical, and content knowledge (TPCK) for educators in the Molecular CaseNet. The report conceptualizes an interdisciplinary framework for teaching about the molecular world and informs lesson design and education research.

Keywords: Molecular education, Case studies, Technological pedagogical and content knowledge (TPCK), Molecular structure and function, molecular visualization, Bioinformatics education, conceptual modeling

Received: 31 Jan 2024; Accepted: 23 Apr 2024.

Copyright: © 2024 Trujillo and Dutta. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

* Correspondence: Prof. Caleb M. Trujillo, University of Washington Bothell, School of Interdisciplinary Arts and Sciences, Bothell, United States

People also looked at