15 Things to include in a thesis proposal

Indicate your research question.
Discuss why this research question is important and/or interesting.
This fourth part doesn’t necessarily go into the thesis, but, for the sake of the committee, explain the method that you used to conduct your literature search to check whether someone has done the exact thing that you plan to do.
Discuss your theory, and provide citations for that.
Indicate the specific hypothesis that you will be testing, which should follow logically from your theory.
Explain, in as much detail as possible, how you plan to test your hypothesis, such as datasets (e.g., name, year collected, sample characteristics, population, if applicable), measurement of outcome variable(s) and your planned coding (such as on a 0-to-1 scale), measurement of the key predictor(s) and your planned coding, measurement of controls and your planned coding, estimation technique (e.g., OLS linear regression, logit regression), and any other details that you can provide at this point, such as whether there would be nontrivial missing data and, if so, your handling of missing data. Provide a sense of your main analysis’s sample size, which – due to missing data – is not necessarily the full sample size of the dataset. Also, provide a sense of the distributions of the outcome variable(s) and your main predictor(s), so that the committee can assess whether the research design is appropriate.
Discuss any robustness checks that you plan to conduct, to provide a sense of the extent to which inferences depend on variation in reasonable research design choices.
Discuss potential limitations of your analysis, such as whether the research design cannot identify causes for certain or whether important measures might be flawed, such as if you were predicting self-reported behavior instead of actual behavior, or if you were predicting a socially desirable attitude in which respondents plausibly would not provide a true socially undesirable response.
Discuss the implications of your analysis, such as whether the analysis would be of interest or use to policymakers or the public or academics. The importance of the analysis should not depend on the results: the key is that others should be interested in your analysis.
Provide references for any citations in your proposal.

15.1 Assessing research report ideas

Is your research question important and/or interesting?
Do you have data and a research design that permit you to plausibly answer your research question? For example, if your research question is about a causal relationship, do your data and research design permit you to plausibly isolate the effect of your key predictor? If your research design is correlational, do you have sufficient controls to isolate the effect of the key predictor and is reverse causality implausible?
Do you have data and a research design such that even a null result would be informative and interesting?
Does your research design have policy implications? For example, if your outcome is a good thing, and your key predictor is manipulable, then a policy implication is that policy should try to increase the level of the key predictor in the population.

15.2 Sample quantitative analysis workflow, for existing data

Determine the proper model – such as Y = X1(educ) + X2(age) – and proper estimation technique (e.g., OLS, logit, ologit).
Clean the data (e.g., recoding, labeling).
Understand the data (e.g., histograms, summary stats).
Revise any codings that should be revised based on the data (such as intended categories with too few observations for statistical analysis, or variables that were intended to be put into a summary index that do not load into the same factor as other variables intended for the index).
Conduct the statistical analysis.
Communicate the key result(s) and the robustness of the key result(s) in visualizations, data tables, and text descriptions.
Clean and label your analysis code and other replication materials for potential readers of the replication materials.