Chapter 3 The Research Question
A good paper starts with a good research question. But how to identify a great research question? In this Module, I will introduce some of the characteristics that make a question a good research question. This section draws on Nick Huntington-Klein’s The Effect.
3.1 Elements to Consider
A great research question needs to fulfill a set of characteristics:
- It is based on theory: Before even having data collected on this, you need to have a good story behind the potential outcomes you obtain. There must be something interesting to say about what you did.
- It is feasible: This characteristic is relevant for different reasons. First, you need to know that the data you need to answer the question is available, or that you can collect it. Second, even if it is possible to obtain the data, you need to be certain that the data is actually going to help you answer the question you have. Third, time is limited, so you also need to consider if you can answer the question within the time frame you considered. Finally, you also need to consider if there is a research design that will allow you to answer your research question. This particularly happens with questions that are more convoluted.
- It is relevant: In the past, I was advised to focus on questions that affect at least a 3% of the world population. Although I believe that number is arbitrary and not necessarily useful, I think that the main point remains. Choose questions that are relevant for at least one community (Los Angeles, San Bernardino, a set of small firms in one region, etc). Of course, projects that address issues that affect a larger share of the population are sometimes easier to address, as data tends to be more available, but you also need to consider novelty, which is my next point.
- It has not been answered before: This point requires some clarification. Researchers build upon each other’s work. We do not reinvent a topic each time. Knowledge progresses at a marginal rate. To do that, research projects do one of three things:
- Contradict a current theory.
- Extend a theory to a new sample to reach external validity (or at least suggest internal validity in a new context).
- Use a different method to test a theory. This is, you do not need to reinvent the wheel with your research project, but need to be original in how you answer the question.
3.2 Identifying Questions
The first step to find your research topic is to think deeply about the issues you care about. These issues, of course, need to satisfy the previous characteristics I introduced. For instance, you may care about your dog Fluffy, and you may want to know what makes him so adorable, but that is not (a) a relevant question for a large share of the population; (b) answerable (you do not have a large sample).
One way to identify potential research topics is by thinking the topics you analyzed in your courses and that can be addressed in an economic way. Think about the courses you liked the most, or a particular topic in a class. This may actually help since probably you already identified some literature on the topic. But remember, the question needs to be analyzed with economic lenses, and needs to fulfill the characteristics that I mentioned previously.
Another way to identify a question is to link it to your future (aspired) career. This, of course, requires that you have a clear idea on what you would like to do in the future, and also needs for you to be original in the way you tackle a subject.
This leads me to my next point. It is hard to identify questions if you do not know what is available, what has been done, and how. This is why you need to read and be curious. You need a critical mind. Just as artists do, obtain inspiration from different, and sometimes unusual, sources. When I ask you to read, this is not limited to academic literature. In fact, a large proportion of what I read is non-academic. Of course we all have a bias on what we read and see based on our interests, but try to broaden your spectrum, you may be surprised about how many interesting things exist out there.
3.3 Determinant Questions
Sometimes, we may be interested in a topic and want to learn and respond everything about it. We get so excited about what are the causes behind a certain issue, that we focus on everything at once.
However, as Marc Bellemare says it, good social science follows the scientific method. This is, you need to have a hypothesis “(x causes Y”), then test it, and finally analyze it.
Determinant papers do not follow this method. Instead, you use a bunch of variables (you have many x’s), but you lack a research design, and therefore just look at correlations, without a clear argument behind it. This type of papers may be useful in some cases (you can read another blog entry from Marc Bellemare on this, or look at the Simpson Paradox on Wikipedia).
You may decide to later on follow a given research agenda, but for one research paper (and particularly, for your senior thesis), keep your question narrow and specific. Briefly, avoid determinant questions!
This also includes avoid giving a clear interpretation to control variables and use them as secondary research questions. Instead, you may want to use a subsample of your data, or disentangle some of the effects behind your main variable of interest.
3.4 Some Examples 2
| Not-so-Good Research Question | Good Research Question |
|---|---|
| What are the determinants of migration? | How does climate change affect migration? |
| What causes criminal behavior? | How does access to public health affect criminal behavior? |
| How does business competition work? | How do incumbent companies respond when a new company enters the market? |
| Is education good? | What are the long-run effects of attending prep school? |
| What causes economic growth? | What is the effect of trade liberalization in promiting economic growth? |