Embedded Files
Hi GBS students ☺️
Please note that this is supplementary to the lecture notes from my methodology course, as well as the thesis guideline I provided during our first group meeting. The relative importance of the sections is indicated by the number of asterisks (*) below. Send me MS Word files for feedback, not PDF. This guideline is mainly for macro-level quant analyses.
Your table of contents should look like below. I encourage you to start by creating these (sub)sections in your Word document and gradually fill them in."
1. Abstract
2. Introduction
3. Theoretical Background and Hypotheses
- A) Research Context (if applicable)
- B, C, D...) A couple of sub-sections for your theory/research stream with hypotheses in them
4. Method
- A) Sampling and Data
- B) Operationalization (or Measurement)
- C) Analytical Approach (or Model Specification)
5. Result
- A) Descriptive Statistics and Correlation Matrix
- B) Hypothesis Tests
- C) Additional Analyses (if applicable)
6. Discussion
- A) Theoretical Contributions
- B) Practical Implications (if applicable)
- C) Limitation and Future Research
7. Conclusion
8. Reference
0. Basics
Do your best to keep your paper between 9000 and 12000 words, regardless of the thesis guidelines.
Include only essential content. Do not stretch sentences or paragraphs unnecessarily.
Use abbreviations consistently: once introduced, use them throughout the paper.
Define key concepts the first time they appear.
Your English writing must be deductive/topic-first: you put the topic sentence in the head of each paragraph and support it with additional elements subsequently.
1. Abstract*
Although the thesis guideline mentions Executive Summary, include "Abstract" instead of it since this is not a consulting paper.
The Abstract should be around 100–250 words.
Present 3–5 keywords underneath the Abstract.
2. Introduction***
Remember what I teach in class: Start with the topic paragraph, then address the research gap (a problem statement), followed by how you aim to fill the gap (both theoretically and empirically). Then, include a summary of findings (optional) and a summary of contributions.
Keep your introduction concise. An unnecessarily verbose introduction will easily cost you -0.5 points on your grade.
Don't make sub-sections. Just one introduction.
3. Theoretical Background and Hypotheses** (the mechanism parts***)
You may use multiple subsections.
You may (or may not) include a subsection for your research context (e.g., The Recent Rise of Corruption in Brazil). This part may not directly relate to your model but helps engage the audience by providing relevant background.
Don’t create multiple sub-sections for every theoretical framework you draw on. Instead, select one or two that form the core of your research. Other frameworks can be mentioned briefly within your arguments, without giving them undue emphasis.
Before each hypothesis, explain the mechanism: “Why do you believe this relationship exists?” “What leads X to impact Y?”
Avoid weak logic like: “Study A found a positive relationship. Study B also found a positive relationship. So I suggest a positive relationship!”
Aim for logic like: “Having a higher X motivates firms to commit more misconduct because they perceive the cost of repercussions to be lower. This is due to reduced media scrutiny, as social ties with the media make firms believe coverage will be less critical.”
Don’t feel pressured to provide citations for "every" sentence. You can strengthen your arguments or fill gaps in the causal mechanism using your own words and by creating your own examples.
4. Method***
Explain your sample, then operationalization (measures), and finally your analytical approach (which model you chose and why) — in that order.
1) Sampling and Data
You may begin by showing your conceptual model (See the figure below).
When discussing your sample, mention the data sources. Go into detail later when you explain operationalization.
Clearly state your final sample. If possible, show how your sample size changed as datasets were merged.
2) Operationalization
Start your paragraph by definining your measure (e.g. ROA is measured as net income over total assets), then elaborate on the details.
The best way to defend your measure is always citing previous papers from good journals.
You may give a bit more explanation about the data source: this is to justify and improve the validity of your measure.
Mention key transformations such as log transformation and standardization. You may omit basic data cleaning steps, such as outlier handling via winsorization.
Alternative measures/techniques for your additional analyses might as well be mentioned in the Additional Analysis sub-section in the Result section.
Controls: ideally two sentences for each. One for the measure. The other for why and how it would affect your DV.
3) Analytical Approach
Explain which model(s) you use for the hypothesis tests (e.g. random effects logistic regression)
You may perform the following tests:
1. Heteroskedasticity test
2. Serial correlation (autocorrelation) test
3. Hausman test (if you do panel analyses)
You can solve the first two by using robust standard errors.
You don't need mathmatical formula/equation for your models.
5. Results*
1. Essential components (SEE TABLES BELOW)
a) Correlation matrix
Just one or two line(s) are enough. Mention you present this in Table X.
The main purpose here is simply to demonstrate that multicollinearity is not a serious issue in your data.
If you check the variance inflation factor (VIF), just mention it in one sentence.
b) Descriptive statistics (can be combined with a))
a) and b) might as well be short. Maybe just a paragraph.
For descriptive statistics, the only essential information for each variable is the mean and standard deviation. Anything beyond that is unnecessary.
Optionally, include supplementary data (e.g., maps showing country-wise variation in key variables). This is not required, just illustrative.
c) Main regression analyses
2. Main results (See tables below)
Present results stepwise: start with the control-only model and add variables incrementally.
Use two to three decimal points only.
Include model fit statistics (e.g., R-squared for linear models, AIC/BIC for probability models).
Omit industry and year dummies from the table but indicate their inclusion in the notes.
Use full names for abbreviations, either at the top of the table or in the footnote.
Tables can be presented horizontally or vertically. Place them in the body text, not as appendice.
Design your tables similarly to those found in published papers. DON't just copy and paste the result tables from softwares.
Use figures only if you have significant moderating effects. Even then, they are not required.
3. Explanation of findings
Keep this section VERY brief and avoid repetition.
If results are surprising (e.g., null or opposite to expectations), provide plausible explanations.
Include a subsection for additional analyses, if applicable. Prepare a separate table(s) for your additional analyses (See the final table below).
6. Discussion**
Begin by reiterating your research motivation, summarizing the gap and objectives, and briefly highlighting key findings. DO NOT just repeat your findings after this.
State clearly which bodies of literature (2–3) you contribute to.
For each, explain what new insight you provide and how your findings link back to that literature.
Always ask: “How should people think differently about this phenomenon after reading my paper?”
Add a Limitation and Future Research subsection after discussing your contributions.
7. Conclusion*
Write a short paragraph that summarizes your key message. This section is not particularly important, so keep it brief.
*The example figures and tables are from my own research projects.
Page updated
Google Sites
Report abuse