Theses, Not Feces

Thesis Writing Workshop

Cuz Potter

College of International Studies, Korea University

2026-05-07

Outline

• Beginning
• AI and theses
• Finding a Question
• Thesis Structure
• Useful Tools
• Conclusion

Resources for this presentation

https://www.cuzproduces.com/producinganew/contents/thesis.php

Beginning

The only good thesis is a finished thesis.

thesis ≠ masterpiece

Download thesis guidelines

AI and theses

Employment and AI

• AI is invading the workplace
• Organizations will need someone to:
  • take accountability
  • write good prompts
• Those people need to understand the content

AI is invading the workplace and will be a component of the work you do, but companies will still need someone to be accountable for failures and someone to write good prompts. To understand whether the AI is hallucinating, you need to understand the material. To write good prompts, you have to understand the material. You can only understand the material if you learn the material by struggling with it.

Theses and AI

• thesis = learning exercise
  
• learning = thinking
  
• thinking = working/struggling
  
• AI → !struggling → !thinking → !learning

What are you going to learn?

Writing a thesis, like all work toward your degree, is a learning exercise. It helps you develop a range of skills, including cold calling, interviewing, problem identification, data analysis, writing, and communication.

Learning leads to employment.

AI is invading the workplace and will be a component of the work you do, but companies will still need someone to be accountable for failures and someone to write good prompts. To understand whether the AI is hallucinating, you need to understand the material. To write good prompts, you have to understand the material. You can only understand the material if you learn the material by struglling with it.

As the presentation goes forward, specific techniques will be shared.

As you will see, however, there is serious disagreement over what is acceptable.

Writing in general

Writing in general

Writing in general

Prompts matter

• Better prompts → better results
• AI can help bootstrap better prompts

Prompts

Develop more elaborate prompts and system prompts to generate better quality feedback. For example:

“You are a university professor specializing in macroeconomics. You are reviewing the attached Master’s degree thesis. The goal of the thesis is to examine ________ using _______ method. What methodological critiques would you make?”

System prompt

System prompts can be created using personalities in ChatGPT or projects in Claude.
System prompts can save many tokens.

Sample system prompt

You are an expert academic research assistant supporting a master’s student developing a thesis in [field/topic area].

Student profile:

• Program: [e.g., International Commerce Track at the Graduate School of International Studies]
• Thesis topic: [e.g., digital public infrastructure in Southeast Asia]
• Theoretical framework: [e.g., world-systems theory, actor-network theory]
• Stage of work: [e.g., literature review / argument development / drafting]

Your role:

• Help develop and stress-test arguments
• Suggest relevant literature and theorists
• Give honest critical feedback — do not simply validate
• Help structure chapters, outlines, and transitions
• Flag logical gaps, unsupported claims, or weak evidence

Output preferences:

• Use academic register but keep explanations clear
• Prefer prose over bullet points in substantive responses
• Prefer bullet points over prose in proposing changes
• When suggesting sources, note why they are relevant to the argument
• Flag when you are uncertain about a citation or claim

Constraints:

• Do not write thesis sections on the student’s behalf
• Always ask clarifying questions before making major structural suggestions
• Prioritize the student’s own theoretical commitments when offering alternatives

Please confirm you understand the context above. Then ask me what I would like to work on today.

Finding a Question

thesis ≠ report

thesis ≠ policy

• A thesis is not a report about a topic or a case. It attempts to add to our understanding of an unresolved theoretical or empirical issue.
• A thesis does not make policies, plans, or programmes. It may make suggestions for how policies can be improved or what should be considered when making policies, but it is not about formulating a policy, plan, or programme.

Follow your interests

…or your business plan

• You’re going to have to put a lot of time and effort into your reading and research. And no matter what you choose, there will be times when you simply want to forget about the whole topic. So make sure you at least like it in the beginning!
• Of course, you may also want to choose a topic that positions you for jobs you would like to get later. For your own sake, I hope those positions are ones you like.
• You may want to tailor the specific topic to market demand, but the overall topic should be of interest to you.
• If you’re not sure what topics interest you, try to be sensitive to your brain’s signals. When you come across a topic of interest, your brain will get excited, perhaps somewhere off in the distance. Learn to listen to your instincts.

Three ways to advance knowledge

(From 김정호교수님)

1. Ask a new question
2. Employ a new methodology
3. Employ a new data set

As 김정호교수님 has said (roughly), “There are three ways you can advance knowledge: 1. Asking a new question; 2. Employing a new method; and 3. Employing a new data set.” A master’s thesis tends to contribute by using a new method or data set. A new method is not necessarily a “new” method; it is one that has not been applied to the particular question or data set that you have identified. It may even be as simple as adding a variable to a regression or using a different type of regression. A new data set could be one that you have downloaded or—quite often at GSIS—a new case study. Researchers may have studied the impact of latrines in reducing water-borne diseases all over Africa, but perhaps they have not done so in Malawi or in the Gesu District of Tanzania.

question
≠
search for policy recommendations

• A question contributes to our knowledge; it does not consist of developing a specific policy.
• Answering a good question may very well have policy recommendations, but these are incidental to the real contribution to knowledge.

Bad sample proposal

Topic: Responsibility to Protect and Non-Interference in Humanitarian Crises: A Case Study

Tentative table of contents:

1. Introduction
   1. Concept of Humanitarian Assistance
   2. Operational system at the international level
   3. Operational system at the regional level
2. Literature Review
   1. Concept of Responsibility to Protect
   2. Non-Interference as Regional Norm
   3. Comparative Analysis
3. Case Study
   1. Examples of need for humanitarian assistance
   2. Analysis of the cases using both concepts
   3. Case of cross-border humanitarian assistance
4. Conclusion

• The topic is really just that, a topic. There is no indication that a real question has been identified. And this typically indicates that the author has not yet done sufficient background reading or work on their literature review. Consider (Booth et al. 2003).
• In fact, the outline may ultimately work, but without a clear question it is simply a list of items that the author believes will be applicable. Quite often a number of items in the list may have no relevance to the final topic. However, if the student has put time into these items, she will typically be reluctant to delete them, leading to a bloated and poorly defined thesis.
• So start with a question and figure out what you need to include to examine it.
• Note that this means doing a lot of reading before you establish your question. A good portion of that reading will ultimately make no direct contribution to your thesis except for having led you to the literature that matters for your question.
• For more advanced readings on social science research, see (Abbott 2004) and (Flyvberg 2001).

What makes a good research question? See O’Leary (2018).

Good research questions define:

1. the topic (e.g., gentrification in Vietnam)
2. the nature of the research endeavor (to discover, explore, explain, describe, or compare)
3. the questions you are interested in (what, where, who, how, when, why)
4. constructs and variables (e.g., tenure, income, legal structure, etc.)
5. the expected relationship between variables (e.g., impacts, increases, decreases, relationships, correlations, causes, etc.)

Thinking about questions

What is the mechanism?
A → ? → B

Examples of research questions

Bad: How does gentrification affect Hanoi residents?
Good: Does gentrification in Hanoi increase incomes for existing renters?

Bad: How can we encourage developing country girls to attend school?
Good: Do well-maintained, gender-separated toilets increase girls’ school attendance?

Bad: What is the impact of the #MeToo movement on cosmetic surgery?
Good: Has the #MeToo movement decreased the revenue of cosmetic surgery offices in Korea?

Bad: How has ODA changed since the Global Financial Crisis?
Good: How has increasing nationalism reshaped ODA flows to HIPCs?

Model your thesis on a journal article

This includes thesis structure.

Thesis Structure

Overall outline

Introduction

1. 750–1,000 words
2. Motivate
3. Introduce question
4. Explain significance
5. Provide signposts

• The introduction should introduce your question and its significance (why it is important).
• I like to start with a compelling incident or fact in order to catch the reader’s attention and make the question concrete in their mind. But this is not necessary.
• Typically the last paragraph very briefly tells the reader what will happen in the rest of the paper. I generally think these are pointless, but they are common. Use your own judgment.

Literature Review

1. 2,000–3,000 words
2. Identify and organize
3. ≠ catalog
4. Focus on causal mechanisms (not your case)
5. → hypotheses

• Writing a literature review is basically teaching yourself about your topic of interest.
• The literature review should identify and organize the schools of thought or differences of opinion about your question.
• It is not a catalog describing related literature; it organizes that literature into broad patterns. By doing so, the literature review typically identifies a number of hypotheses to test and variables to include in your analysis. A good literature review, in my opinion, makes the rest of the thesis relatively straightforward: it provides a clear question and framework for addressing it.
• Note that literature reviews generally do not include background on your case study (if you are doing one) or data set. The literature review examines what we know about a broad question that you examine through a specific case or data set.
• You will have to read a lot of articles before you zero in on your specific topic. Therefore, you will undoubtedly read much more than you include in your thesis. Do NOT include ideas or content just because you read and learned about them. Extraneous material reduces the overall quality (and thus assessment). Do NOT cite irrelevant articles just to increase the count.
• Do not be afraid to use older journal articles. Good articles are good, no matter when they were written.
• You should not be afraid of drawing upon literature reviews that are either in the articles you are reading or that have already been written. For writing literature reviews, look at Creswell (2003) or Hart (1998).

AI and the literature review

AI and the literature review

AI and the literature review

Methodology

1. 500–1,500 words
2. Explain techniques
3. Identify data sources
4. Report limitations

• The methodology section describes the techniques and data sources employed in your analysis. It should supply sufficient detail for the reader to understand how you collected your data and how to replicate your study.
• The methodology should also report the limitations of your method as to validity and generalization. That is, you should tell the reader under which conditions your study is relevant and how far he or she can extend the study. For instance, a study of microfinance in the Middle East may not be relevant in areas that do not employ Islamic financing rules.
• For guidelines in developing case study research (and potential language for describing it!), consider looking at Yin (2003), especially chapter 2.

Quantitative versus qualitative

1. Quantitative
   1. Broad
   2. Existing data → easier
   3. New data collection difficult
   4. Employability ↑
2. Qualitative
   1. Detailed
   2. Data collection difficult
3. Triangulate (= mixed-methods)

• There are advantages and disadvantages to any research method.
• Quantitative methods seem simple, but are quite difficult to do well. In particular, accessing sufficient data to conduct meaningful research can be quite difficult. Collecting your own is exceptionally challenging. Using existing data sets often means changing your question to suit the variables available in your data set. That said, the demonstrated ability to conduct quantitative analysis can contribute to your employability.
• Qualitative methods require a different intensity of involvement with the subject of your research and can thus be more time consuming. You can also run into difficulties contacting the people you need to contact and in getting useful information from them. So start early! That said, interviewing can be a great way to build networks that might lead to employment.
• Both approaches can be rigorous. And both can be sloppy!
• Another approach is to triangulate. That is, you use some quantitative analysis and some qualitative analysis to reinforce each other.
• People doing development theses face additional difficulties in conducting qualitative research. If you do not plan to travel to the place you are studying, it can be virtually impossible. You can conduct online interviews (video, voice, email), but this is generally not as robust. You may want to take a look at Desai and Potter (2006) for more ideas.

AI in your methodology

AI in your methodology

AI in your methodology

AI in your methodology

AI in your methodology

Findings

1. 2,500–3,500 words
2. Descriptive, not analytical
3. Divide as necessary (e.g., by hypothesis, case, topic)

• The findings section may be multiple chapters, depending on how you would like to organize it. This section reports the results of your quantitative study or lays out the relevant details of your case study. In some ways, this is the driest part of the thesis as it simply reports your findings and provides foundational interpretations of the results. It does not interpret your findings.
• Do NOT include ideas or content just because you read and learned about them. Extraneous material reduces the overall quality (and thus assessment). This applies particularly to case studies. For example, the reader does not need to know the main industries in the country you are studying to understand the impact of school fees on student attendance.

AI in your findings

AI in your findings

Discussion

1. 1,500–2,000 words
2. Purposes:
   1. Interpret and explain your findings
   2. Answer your research question
   3. Connect findings to literature review
3. Must relate back to the literature review

The discussion section is where you provide an interpretation of your findings and answer your question.

AI in your discussion

Conclusion

1. 750–1,000 words
2. Reintroduce question
3. Quick summary of findings
4. Policy recommendations
5. Implications for literature
6. Future research
7. Revisit validity and generalization?

• The conclusion should reintroduce your question, offer a quick summary of your findings, make policy recommendations, and (often) make recommendations for further research.
• You may also want to review the limits to your study’s validity and generalization. Some professors like to see this at the end. Personally, I hate to end on a negative note.

Other parts of the thesis

1. Cover
2. Abstract
3. Acknowledgments
4. Table of contents
5. List of tables
6. List of figures
7. Bibliography
8. Appendices
   1. Questionnaires
   2. List of interviews
   3. Additional tables or diagrams of potential interest

Abstract

1. Roughly 250 words in single paragraph
2. abstract ≠ introduction
3. Contains:
   1. Research purpose/question
   2. Methodology
   3. Major findings
   4. Conclusions

Useful Tools

Useful resources

1. Google Scholar (https://scholar.google.com)
2. ISI Web of Knowledge
3. Lexis-Nexis
4. JSTOR
5. Access through KU network or by logging in to the library website
6. https://www.cuzproduces.com/producinganew/contents/thesis.php

• Google Scholar can be used in many productive ways. Here are two important tips. First, you can often search for your topic + “literature review” to find a preexisting literature review that you can draw on. Second, Google Scholar tells you how many times an article has been cited. This does not mean that less-cited articles are less useful or that more-cited articles are better. It does mean that the article constitutes a cornerstone for a lot of people interested in the same topic and that you should probably read it first.
• ISI Web of Knowledge allows you to search forward and backward from an article you are interested in. You can find out which subsequent articles cited the original article and which articles it cited. The first allows you to find more contemporary discussion of older pieces.
• My website cuzproduces.com offers copies of the presentation, some of the resources identified here, and other materials that may help you approach your thesis.

Bibliographic databases

1. Endnote
2. Zotero
3. Reference Manager
4. JabRef

• A bibliographic reference manager stores all the information about the pieces you have read in one place. More importantly it formats and generates your bibliography automagically. It can also change the format automatically for you. This will save you a lot of work in the long run.
• Useful quotes from your journal articles go into the bibliographic database. This can help you avoid plagiarism by reducing the tendency to go back to the original article.

Styles and formatting (word processors)

1. Facilitates formatting changes
2. Allows automation of other document parts, like tables of contents

• “Styles and formatting” is an essential word processing technique. Learn how to use it. If you are consistent, you can easily and quickly change the format of any category of your document, e.g., section titles.
• Proper styles and formatting also automates tables of contents, lists of figures and tables, etc.

Conclusion

Summary

1. Choose a topic that interests you.
2. Start with a question, not a topic.
3. Your lit review is the thesis’s foundation.
4. Findings convey content of your research.
5. Discussion connects findings to lit review.
6. Cite properly.

Thank you.

https://www.cuzproduces.com/producinganew/contents/thesis.php

References

Abbott, Andrew. 2004. Methods of Discovery: Heuristics for the Social Sciences. W. W. Norton & Co.

Booth, Wayne C., Gregory G. Colomb, and Joseph M. Williams. 2003. The Craft of Research. The University of Chicago Press.

Creswell, John W. 2003. Research Design: Qualitative, Quantitative, and Mixed Methods Approaches. Second. Sage Publications.

Desai, Vandana, and Robert B. Potter, eds. 2006. Doing Development Research. SAGE Publications.

Fainstein, Susan S. 2001. The City Builders: Property Development in New York and London, 1980-2000. 2nd ed. The City Builders. Property Development in New York and London, 1980-2000. Studies in Government and Public Policy. University of Kansas Press.

Flyvberg, Bent. 2001. Making Social Science Matter: Why Social Inquiry Fails and How It Can Succeed Again. Cambridge University Press.

Hart, Chris. 1998. Doing a Literature Review: Releasing the Social Science Research Imagination. Sage Publications.

O’Leary, Zina. 2018. Research Question. SAGE Publications Ltd. https://www.ebook.de/de/product/33315335/zina_o_leary_research_question.html.

Yin, Robert K. 2003. Case Study Research : Design and Methods. 3rd ed. Applied Social Research Methods Series ; v. 5. Sage Publications.

Plagiarism

Stealing is bad, m’kay?

GSIS definition of plagiarism

Plagiarism is the act of presenting the creative work of another person as if it were your own, without appropriate acknowledgment of the author or source.

— {visibility=“uncounted”}

Nothing is original. Steal from anywhere that resonates with inspiration or fuels your imagination. Devour old films, new films, music, books, paintings,… Select only things to steal from that speak directly to your soul. If you do this, your work (and theft) will be authentic. Authenticity is invaluable; originality is non-existent. And don’t bother concealing your thievery—celebrate it if you feel like it. In any case, always remember what Jean-Luc Godard said: “It’s not where you take things from—it’s where you take them to.”

—Jim Jarmusch

• Celebrate = show → “Celebrate it if you feel like it” → “Document your theft”
• Our ideas are not original. The way we combine them can be. And it is the combination that is the contribution.
• Give credit where credit is due.

— {visibility=“uncounted”}

How do you avoid plagiarism?

— {visibility=“uncounted”}

CITE!!!

오빠, Chicago style

Chi-, Chi-, Chi-, Chi-, Chicago style!

Chicago style in text

1. “Major developers found themselves in increasing difficulty” (Fainstein 2001).
   • Citation comes after all cited material so that the reader does not confuse others’ ideas with yours.
2. As Fainstein (2001) argues, “Major developers found themselves in increasing difficulty.”
   • Implies that following material is drawn from another source, until you add a clear transition, like, “Contrary to her argument…”
3. As Fainstein argues, “Major developers found themselves in increasing difficulty” (Fainstein 2001).
4. “Major developers found themselves in increasing difficulty” (Fainstein 2001, 44).
5. Blah, blah, blah (Abbott 2004; Fainstein 2001).

• GSIS uses Chicago author-date style, so start using it in ALL your papers.
• There are many online resources that detail the style. Use them.
• Better, use a bibliographic reference manager.
• It may seem awkward at first, but for the informed reader, the author-year style indicates quickly where you have obtained your ideas and understanding.

Types of plagiarism

Four types of plagiarism¹

1. Verbatim (Ctrl-C, Ctrl-V)
2. Inadequate paraphrase (Remixes and mashups)
3. Uncited paraphrase (IP theft)
4. Uncited quotations (Duh!)

1. This section draws heavily on the Harvard Guide to Using Sources and plagiarism.org.

Ctrl-C, Ctrl-V

Source

The steeply climbing curve of returns from real-estate investment prompted a stream of new development proposals, which justified their costs with prognoses of ever-increasing returns.

(Fainstein 2001)

Plagiarized?

Real estate investment increased dramatically in the 1990s because the steeply climbing curve of returns from real-estate investment prompted a stream of new development proposals, which justified their costs with prognoses of ever-increasing returns.

Ctrl-C, Ctrl-V: Solution 1

Source

The steeply climbing curve of returns from real-estate investment prompted a stream of new development proposals, which justified their costs with prognoses of ever-increasing returns.

(Fainstein 2001)

Solution 1 = direct quotation with citation

Real estate investment increased dramatically in the 1990s. As Fainstein (2001) suggests, ”the steeply climbing curve of returns from real-estate investment prompted a stream of new development proposals, which justified their costs with prognoses of ever-increasing returns.”

Ctrl-C, Ctrl-V: Solution 2

Source

The steeply climbing curve of returns from real-estate investment prompted a stream of new development proposals, which justified their costs with prognoses of ever-increasing returns.

(Fainstein 2001)

Solution 2 = paraphrase with citation

Real estate investment increased dramatically in the 1990s. As Fainstein (2001) suggests, increasing returns on real estate investment were used to justify new proposals.

Remixes and mashups

Source

The steeply climbing curve of returns from real-estate investment prompted a stream of new development proposals, which justified their costs with prognoses of ever-increasing returns.

(Fainstein 2001)

Plagiarized?

Real estate investment increased dramatically in the 1990s. As real estate investment returns increased steeply, a stream of new development proposals justified their costs with predictions of ever-increasing returns.

• Changes the order of the sentence but retains phrasing.
• Replaces a few words (highlighted) with synonyms.
• Would be the same if multiple sources were combined.

Remixes and mashups: Solution 1

Source

The steeply climbing curve of returns from real-estate investment prompted a stream of new development proposals, which justified their costs with prognoses of ever-increasing returns.

(Fainstein 2001)

Solution 1 = adequate paraphrase with citation

Real estate investment increased dramatically in the 1990s. In the wake of increasing returns from developments earlier in this period, developers used these returns to justify new proposals (Fainstein 2001).

• Despite a few identical or nearly identical words, phrases are not copied.
• Author is cited.

Remixes and mashups: Solution 2

Source

The steeply climbing curve of returns from real-estate investment prompted a stream of new development proposals, which justified their costs with prognoses of ever-increasing returns.

(Fainstein 2001)

Solution 2 = direct quotation with citation

Real estate investment increased dramatically in the 1990s. Building on increasing returns from developments earlier in this period, new proposals “justified their costs with prognoses of ever-increasing returns” (Fainstein 2001).

• Directly quoted phrases are quoted.
• Author is cited.

IP theft

Source

The steeply climbing curve of returns from real-estate investment prompted a stream of new development proposals, which justified their costs with prognoses of ever-increasing returns.

(Fainstein 2001)

Plagiarized?

Real estate investment increased dramatically in the 1990s. In the wake of increasing returns from developments earlier in this period, developers used these returns to justify new proposals.

IP theft: Solution

Source

The steeply climbing curve of returns from real-estate investment prompted a stream of new development proposals, which justified their costs with prognoses of ever-increasing returns.

(Fainstein 2001)

Solution = adequate paraphrase with citation

Real estate investment increased dramatically in the 1990s. In the wake of increasing returns from developments earlier in this period, developers used these returns to justify new proposals (Fainstein 2001).

or

Real estate investment increased dramatically in the 1990s. According to Fainstein (2001), in the wake of increasing returns from developments earlier in this period, developers used these returns to justify new proposals.

• Author is cited.

Duh!

Source

“The steeply climbing curve of returns from real-estate investment prompted a stream of new development proposals, which justified their costs with prognoses of ever-increasing returns.”

(Fainstein 2001)

Plagiarized?

Real estate investment increased dramatically in the 1990s. As a result, “the steeply climbing curve of returns from real-estate investment prompted a stream of new development proposals, which justified their costs with prognoses of ever-increasing returns”. And politicians failed to see the false logic this entailed.

Duh!: Solution

Source

The steeply climbing curve of returns from real-estate investment prompted a stream of new development proposals, which justified their costs with prognoses of ever-increasing returns.

(Fainstein 2001)

Solution = quotation with citation

Real estate investment increased dramatically in the 1990s. As a result, “the steeply climbing curve of returns from real-estate investment prompted a stream of new development proposals, which justified their costs with prognoses of ever-increasing returns” Fainstein (2001). And politicians failed to see the false logic this entailed.

• Author is cited.

In practice

Build your ideas into an outline.

Then write.