Research Design and Methods
A research proposal must discuss both the research design and the research methods. In doing so, your goal is to outline a systematic plan that is convincing to you, your advisor, funding agencies and anyone else reading your proposal. There are, thus, two key descriptors of your plan; it should be both systematic and convincing. First, you want to ensure that you take a systematic approach to your research. Doing so is important to guaranteeing that you collect the evidence necessary to support your argument and argue against competing hypotheses. By approaching your research in a systematic manner, you help achieve the second goal: convincing yourself and your reader. Specifically, being systematic helps to convince your reader that you have sufficient evidence and that what you present, both in favor of your argument and against competing hypotheses, is representative of the universe of evidence and was not cherry picked to support your point.
There are three broad questions you need to address when writing about design and methods:
Cases
In thinking about your design and methods, the first question you’ll want to answer is what cases you plan on including. The answer here could range from all 195 countries in the world to the town of Buford, Wyoming (the smallest town in the United States). In telling your reader about your case selection you should consider discussing the following questions:
What is the correct unit of analysis for your question?
National governments? Subnational governments? Bureaucratic agencies? Individual voters?
How many cases will your research project cover?
Why did you choose the number of cases that you did? To justify this decision, discuss the trade-offs between depth and breadth of analysis and why your question is best answered by the balance you chose.
What analytical leverage do you gain from your case selection? In this discussion consider the following:
Based on what you know, is your hypothesis likely or unlikely to be true in your cases? You might hear this referred to as whether your case is most likely or least likely. What does this tell you about what you might expect to see in other cases?
If you have more than one case, how do the cases compare? What do they have in common? What are the relevant differences? Focus on similarities and differences that relate to the variables in your theory, as well as competing explanations that you want to refute.
When you are working with a small number of cases, it can be helpful to represent this information in a table, which allows for an easy, visual comparison of the cases on the relevant dimensions. For a large number of cases, a table may still be helpful. However, you may consider reporting the ranges of the relevant variables, rather than having a column (row) for each case.
What do these comparisons allow you to say about your question? Do the similarities help you control for possible competing explanations? (this is known as the method of similarity) Do their differences help you to identify the relevant independent variables? (this is known as the method of difference) Do they show different mechanisms?
What is the universe of possible cases?
The universe of possible cases is limited in part by your unit of analysis.
It may be limited even further by the scope conditions you have placed on your hypotheses. While you have likely already discussed these scope conditions earlier in your proposal, it is important to remind your reader of them as part of your discussion.
What are the specific cases you chose?
Are there any logistical justifications (such as linguistic or cultural knowledge or relationships with people who can help overcome administrative barriers) for your case selection?
Logistical factors should not be the only reason for case selection. However, since you cannot learn anything from a research design that you cannot execute, it is important to also consider logistical factors.
Empirical Strategy
After discussing your case selection, the next question you’ll want to answer is what techniques you will employ to collect and analyze evidence from your selected cases.
In this section of the proposal, you should describe your strategy and defend why it is the best one to answer your research question. In this discussion, consider the following questions:
What type of evidence do you need to support your theory? Why is the empirical strategy you propose the best for providing this type of evidence?
Are there other empirical strategies that a reasonable person could propose to provide this evidence? Why is your chosen strategy superior?
What are the weaknesses of your strategy? Why, despite these weaknesses, is your approach the best one?
Depending on the audience for your research proposal, you may also want to discuss the preparation and skills you have (such as knowledge of regression models or a foreign language) that allow you to implement this specific empirical strategy. This is particularly important when sending your proposal to funding agencies, since you are trying to convince them not only that your research question is worth answering, but also that you are the person who should carry out this research.
Some of the information you should provide is different for each empirical strategy. Below is a list of questions for some of the strategies most commonly utilized in political science and by Government undergraduate thesis writers.
Statistical Analysis
What database(s) will you use? Where are the data available from?
What variables will you use?
How will you address the problems of omitted variable bias and reverse causality?
What will your main regression equation be?
Archival Work
What documents will you examine? How can you access these documents?
Will you extract qualitative or quantitative data from these documents?
Are the documents in the archive a complete representation of what was written at the time? If not, how will you address this potential bias?
Interviews
Who will you interview? The “who” in this question refers to categories of people, rather than the names of specific individuals. If there are significantly more people who fit your criteria than you plan to interview, discuss how you will limit your sample.
What types of questions will you ask your interview subjects? What topics will you discuss?
How will you gain access to interview subjects?
Process Tracing
Will you use process tracing to draw descriptive or causal inference?
What is the source of the information that you will use to conduct process tracing?
What type of evidence do you expect would constitute a straw-in-the-wind test? A hoop test? A smoking gun test? A doubly decisive test?
Content Analysis
Will you use qualitative or quantitative (automated) methods of content analysis?
What documents will you examine? How can you access these documents?
What will you look for within these documents?
Mixed Methods
Many research projects involve the use of mixed methods, or a combination of more than one empirical strategy. A research proposal for a project that uses mixed methods should discuss each of the methods individually, answering all of the relevant questions outlined above. However, it should also discuss the role that each method will play in the overall strategy. There are a number of roles that methods can play. For example, you may use one method to generate theory and another to test theory, or you may use one method to provide evidence of the relationship between the independent and dependent variables and a second to illustrate the mechanism that connects these variables.
Measurement
Example
Independent Variable: democracy
Definition: free and fair elections (note that this is a simplified, and thus incomplete, definition)
Grouping (Option #1): Binary, countries are democratic or they are not
Grouping (Option #2): Continuous, countries are more or less democratic, but there is no clear cut-off between democratic and non-democratic countries
Observable Aspects (Option #1): Percentage of the adult population that participates in elections. Since democracy is a binary variable, countries where greater than a certain percentage of the population votes are democratic and those below that threshold are not.
Observable Aspects (Option #2): Percentage of the adult population that participates in elections. Since democracy is a continuous variable, countries with a larger percentage of the population voting are more democratic than those with a smaller percentage.
Finally, your proposal should discuss how you will measure the relevant variables for your project. In answering this question you can describe the measurement not only of your independent and dependent variables, but also of your mechanism and the variables from key rival hypotheses.
Writing about measurement, which you can think of as describing how you will know each of your concepts when you see it, requires answering two questions:
What are the groupings of your variable? Is it binary? Categorical? Continuous?
What are the observable aspects of your definition?
Since you will have to answer these questions for a number of variables, a table summarizing the answers to these two questions for each variable can be a helpful complement to the prose.
Tying it Together
Below is a sample of writing about case selection. The video on the right walks through the role that each sentence plays in answering the various questions outlined above. The video also discusses how to “justify your choices” when writing a research proposal.
Research Question: Does democratization cause higher levels of educational attainment?
Theory: Democracy leads to Increased responsiveness leads to Increased educational spending leads to Increased educational attainment
Case Selection:
Given that regime type is a country-level variable, I plan to study the relationship between my independent and dependent variable at the national level. The mechanism that I propose stems from the behaviors of individuals, meaning that the evidence towards my mechanism should come from the study of individuals. As I do not have any scope conditions for my theory, I propose to include all countries in my analysis of the relationship between the independent and dependent variable. This approach allows me to speak to the generalizability of this relationship, more so than would be possible with a narrower set of cases. While qualitative methods might provide a more nuanced picture of both democracy and educational attainment for a given country, they are not necessary to answer my research question. What is most important to my question is documenting the presence of democracy, as well as the number of years of schooling that the average student completes. Both can be easily documented quantitatively, thus allowing me to accurately collect evidence towards my question for most countries in the world.
The depth of analysis needed to show my proposed mechanism means it would be difficult to provide this evidence for all countries. Therefore, I propose to explore individual behavior regarding electoral pressure and public education in two countries: Finland and Mexico. Both of these countries are democratic, but have few other similarities, which helps me show that democracy is what drives changes in educational provision. In each country I will collect evidence about average voters’ beliefs and behaviors regarding education. Finland is a most likely case for my mechanism, since electoral accountability is strong and the average citizen is considered to be informed about education policy. Mexico is a least likely case for showing my mechanism, given the ways in which clientelism perverts electoral accountability. Therefore, if I observe evidence of my mechanism in Mexico, I will conclude that it is likely that my mechanism plays out in most countries. Similarly, if I do not observe evidence of my mechanism in Finland, I will conclude that it is likely that my mechanism does not occur in most countries.
By combining these two sets of cases, I can provide the breadth of evidence necessary to speak to the generalizability of the relationship between my independent and dependent variables, as well as the depth of evidence necessary to speak to my mechanism. While this approach does not allow me to document the mechanism as generalizable of a manner, by choosing to explore my mechanism in both a most-likely and a least-likely case, I can also make informed conjectures about the generalizability of my mechanism outside of those two cases.