Topic, question, thesis
- Make sure your paper is analytic. Argue for an analytic (or “positive”) stance, rather than a normative one. In other words, engage in “objective” rather than “subjective” analysis. Answer the question “what is?”, rather than the question “what should be?”.
Audience and voice
- You are writing for a professional audience. Choose language and a tone that reflect this.
- Provide enough political, social, and/or intellectual context for your reader to understand the issue you address, and why one should care about it.
- When describing procedures you undertook, write in the first person. For example, write “We collected data from 1000 web sites” or “I then regress the outcome on two predictors,
X1 and X2”. Avoid the passive voice. (Do not write “The data from 1000 web sites were collected” or “The outcome was regressed on X1 and X2”. These leave us wondering, “by whom?”)
Research strategies
- Clearly state your specific question. Your question is a case of a more general set of political phenomena. Know the literature from the more general set.
- Compare your political phenomenon of interest to models of how politics works in such cases.
- Understand counterarguments, alternative explanations, and competing hypotheses to your own. Refute them to the extent possible.
Content
- Be clear whether you’re making descriptive or causal claims.
- Make claims that are falsifiable. How would you know if you were wrong? What evidence would disagree with your argument?
- In making causal claims, the nature of the implicit or explicit comparison group is important.
- In making causal claims, do not control for or match on post-treatment quantities; avoid post-treatment bias.
- Discuss measurement, modeling, design, and data strategies. Discuss limitations of your approaches.
- Do not include the specific names of variables or objects in your paper. It can be important to describe a variable you created, but its name is not important to the reader. For example, you might write “We create a binary measure of ideology that is 1 for respondents with scores greater than the median and 0 for those at or below the median.” You should not include the fact that you named the variable
ideo7gr_med.
- Provide the details of your quantitative results.
- Web sources should exclude wikis, include full citations (including access date), and be kept to a minimum. When possible, cite the original published document instead of the web version.
Style
- Be clear.
- Be concise.
- Be coherent.
Further advice
- On every step of the research process (data, coding, writing, presentation, management), see Harvard IQSS’s Undergraduate Research Resource Reservoir.
- On writing nonfiction, read William K. Zinsser’s On Writing Well (noting that academic writing is different than humor writing, e.g.).
- On writing for publication, read Gary King’s online notes and PS: Political Science and Politics article (PDF).
- On writing for publication, read Andrew Gelman’s suggestions.
- On citation, see the APSA citation style, the Chicago Manual of Style, and the MLA Handbook.