Research Proposal Assignment 2021-2022
POLS0012: Causal Analysis in Data Science
This document outlines the instructions for the research proposal component of the assessment for POLS0012
in 2021-2022. The research proposal that you write for the course will constitute 50% of your overall grade
and will be due on 12th January, 2022 at 2pm.
The other 50% of your grade for this module will come from answering a set of quantitative questions which
will be released at the end of term. That part of the assessment will also be due on 12th January, 2022 at
2pm.
Research Proposal
Your task is to design a research paper that answers a causal question. Present your design in a written
research proposal that outlines how you would carry out your research, justifies your proposed methodology,
and explains its potential limitations. You should design a single research paper, like the examples you
encountered in this module. Your proposed paper can address any issue of your choice from economics,
human geography, political science, public health or public policy, provided you are asking a clear causal
question. You are free to propose any method (or combination of methods) to answer it, on three conditions:
1. You must propose using at least one of the techniques of causal inference covered in this mod-
ule: experiments, matching, regression, instrumental variables, regression discontinuity, difference-in-
differences/fixed effects, or synthetic control. For example, you could design a field experiment, or
propose a suitable instrumental variable, or a setting for an RDD.
2. You must not use your undergraduate dissertation topic for this question, in order to ensure that
the workload for this assignment is not disproportionately small for those students whose dissertations
involve using a technique of causal analysis.
3. You must not plagiarise an existing study, published or not. It is perfectly acceptable for your proposal
to be inspired by an existing study, but if this is the case you must say so, citing the existing study.
Your proposal should be achievable and realistic. The data that you require should exist, or should be
collectible in principle. However, you may assume that you have a large research budget, so that it is feasible
to carry out a project that is expensive or logistically complex, where necessary.
Your essay should consist of a written research proposal. It must contain details of:
• The causal question you have chosen to answer.
• The data that would be used to carry out your study.
• A technique (or set of techniques) of causal analysis that is designed to answer your question, and an
explanation of why it is appropriate.
• The statistical analysis that you will carry out.
• Any potential limitations of your approach: what assumptions are needed for your study to yield a
causal effect? How valid will your results be?
• Any robustness checks or other exercises that you will carry out to help mitigate these potential
concerns.
1
Please note the following guidance:
• This is a written task only. No statistical analysis should be included.
• Your proposal can be structured in sections with sub-headings.
• Choose a research question that is clearly defined. Outline concrete steps that you would take to answer
it.
• Do not worry about choosing a research design that is perfect. Such a thing almost certainly doesn’t
exist. Instead, show that you have thought carefully about the strengths and drawbacks of your
proposal.
• Higher marks will be reserved for proposals that are original, i.e. not a minor variation on an existing
study. I recommend thinking of a research question that interests you and designing a study from there,
rather than taking an existing paper and trying to tweak it for this essay.
• Higher marks will also be reserved for proposals that would be feasible, e.g. clearly laying out where
the data will come from.
