Crafting research questions should be fun. Okay, we can’t lie: it may not always be fun, but it should be at least interesting. As advisers, though, we have found that it is rarely an easy task to elicit “good” questions from students. In fact, when we ask students who are beginning the process what questions they are interested in, we almost always get some version of “How does a {generic concept} affect some {broad other thing}?” Some classics in political science include questions like “Does religion affect politics?” or “How does public opinion affect trade policy?” or “How does ideology make us dumber?” (Okay, that last one may just reflect Nick’s id.)
These are perfectly acceptable questions to ask. In fact, entire fields develop around them! But they are not research questions. That is, they do not pose a specific question that could orient a modest project designed to tell us something new about the relationship between two things.
Instead, the trick to finding a research question involves identifying specific aspects of these general topics and linking them together in plausible ways. Put another way: be as specific as possible in enunciating a logic of “cause” and “effect” to orient the research question.
We like to return to the language of empirical social science when teaching research questions. Social scientists are often interested in how some independent variable (sometimes notated as X) affects an outcome or dependent variable (sometimes notated as Y). For example, you might wonder “How are religious beliefs related to death penalty preferences?” or “How does partisanship shape beliefs about taxes?”
These are good starting points - certainly they are more specific than our very vague questions above! Let’s call these types of questions topical questions because they ask about a relationship between two topic areas (e.g. religion and political attitudes) without supplying any sort of mechanistic logic about how these things might be related.
Because we are often interested in whether X causes Y (or, in a weaker descriptive sense, whether X is correlated with Y), we need to shift how we think about conceptualizing the questions we ask. Instead of asking whether two topics are related to each other in a general sense, we need to express a research question in terms of how some feature about our independent variable (X) affects a specific component of our topic of interest, our dependent variable (Y).
Often, in public opinion, we can think about this in terms of whether exposure to an informational treatment, a demographic feature of the respondent, or some other attitude a person holds “increases” or “decreases” favor or opposition to our outcome of interest. In keeping with the religion and death penalty question above, we could narrow this down in many different ways. We could ask,
“Do religious messages about the sanctity of life affect death penalty attitudes?”
“Does listening to sermons about a literal hell affect whether people support the death penalty?” or even
“Do Christians and Muslims differ in their beliefs about the death penalty?”
Here, we’ve re-packaged our general interest in the relationship between religion and politics into more specific research questions that taps into religious content. Maybe we’re interested in the effect of sermons (1,2) or whether different faiths have different views about the death penalty all together (3).
Our examples in this edition have involved public opinion in the context of American politics, so this process has been straightforward. Don’t be discouraged, though: you can repeat this process for any topic. Whatever students are interested in, they must find a way to explore concretely how a specific “thing” affects some other “thing.”
The logic above could just as easily be applied to institutional contexts or other fields entirely. You can imagine, for instance, being interested in whether the democratic peace only holds when other countries believe that each other are democracies or whether it’s really the institutional design of each regime that matters. Similarly, you could imagine asking whether Americans’ opinions toward free trade is affected by how their community (rather than their own job) would be influenced.
The takeaway: A research topic is not a research question.
The practical: Don’t be ambiguous. You don’t want to simply ask “How is X related to Y?” Instead, our research questions should produce a testable prediction involving a directional effect that one thing has on another. For example, in a general form, “Does X decrease support for Y?” If you can generate a hypothesis from the research question, then that is a great sign that you’ve settled on something for which you could spend a semester, a year, a thesis, or a lifetime of study on.