Eastern Michigan University
page banner

Developing a Good Research Question

Developing a good research question is the key to getting your dissertation done efficiently and to making an original contribution to your discipline. Your dissertation question should meet six criteria.

Identify the Theoretical Construct

A good research question should clearly identify the theoretical construct you are studying. For example, if you are interested in figuring out the processes by which parents transmit their political perspectives to their children, the theoretical construct you are studying is "transmission of political perspectives." If you are interested in how technology innovations in teaching improve student performance, your theoretical concept is "effectiveness of innovation." Notice that the theoretical construct is the phenomenon, event, or experience you want to learn more about.

Recognize the Theoretical Construct

A research question should contain some suggestion of recognizability of the theoretical construct. This means that the research question articulates the theoretical construct in a way that is specific enough so you will know it when you see it as you are coding for it in your data. In other words, it supplies a clear unit of analysis that allows you to tell the difference between that construct and other constructs relatively easily. To accomplish recognizability, you should word the construct in a way that is concrete and specific.

Perhaps an example will help clarify this idea of recognizability. A student started her dissertation planning with a theoretical construct of "the experience of nontraditional women in college." While certainly an important construct, it is too large because the student would have a difficult time recognizing the construct in the data. It involves a potentially large number of different constructs, including women's experiences of raising children while going to school, degree of support from family members, responses of other students, educational accomplishments, emotions the women experience, and on and on and on. There is virtually nothing having to do with nontraditional women college students that would not count as part of the construct of "the experience of nontraditional women in college."

A more specific theoretical construct would be "nontraditional women's experiences of discrimination in the classroom" or "nontraditional women's use of support services on campus." The recognizability here is that the theoretical construct is focused on one aspect of nontraditional women's experiences and allows the student to discriminate between it and other constructs that are a part of nontraditional women's experiences in college. As you formulate your research question, think about how you will code data with that question, looking for examples of the theoretical construct you are considering featuring in your research question. Will you be able to locate it and distinguish it easily from other constructs that appear in your data?

Transcendence of Data

Your research question should meet the criterion of transcendence of data. Except in a few instances, your research question should not include mention of the specific data you are using to investigate your question. Many different kinds of data can be used to answer your question so don't confine it to the one type of data you plan to study. You want your question to be more abstract than those specific data.

For example, if you want to study resistance strategies used by marginalized groups to challenge institutions, you can use as your data a social movement, works of art by politically motivated artists, the songs sung by union organizers, or the strategies used by Mexican immigrants to improve their status in the United States, to name a few. You want your study to contribute to a significant theoretical conversation in your field, and it can do that more easily if your question is not tied to one particular kind of data. A research question on the topic of resistance that transcends the data, then, might be, "What is the nature of the resistance strategies used by subordinate groups in their efforts to challenge hegemonic institutions?"

As an example where the criterion of transcendence of data was violated in a research question consider the proposal of a theoretical construct of "accounting practices used in children's theatres in Detroit." Here, a theoretical construct is the same as the data. The student is conflating the construct in the research question with the data used to answer the question. As a result, the story has limited interest to other readers. The students certainly could collect data for a study concerning accounting practices in children's theatre groups in Detroit, but the construct should be larger than that. Perhaps it could be something like "accounting practices in nonprofit arts organizations."

There are a few kinds of dissertation where the criterion of transcendence of data in the research question does not apply. These are dissertations in which researchers want to find out about a particular phenomenon, so the research is specifically about that phenomenon. For example, someone who is interested in the strategies used by Alcoholics Anonymous to attract members would want to include Alcoholics Anonymous in the research question. In this case, the researcher sees something unique and significant about that particular organization, in contrast to other treatment approaches, and sets out to understand it specifically.

There are some fields, too, where the data are typically included in the research question in dissertations. History is one. Dissertations in this field are about a particular place and time, and their purpose is to explore that place and time. Thus, those particulars are included in the theoretical construct of the research question. For example, a research question for a history dissertation might be, "How was a counter-culture identity sustained in Humboldt County, California, in the 1980s and 1990s?" The discipline of English is another one where research questions may include mention of data. Scholars in English are often interested in a writer or group of writers or a particular type of literature, and those would be included in the research question. An example is: "How do troll images function in the narratives of Scandinavian writers between 1960 and 1990?"

Contribute to Understanding the Construct

Your research question should identify your study's contribution to an understanding of the theoretical construct. Your research question should name what happens to the theoretical construct in your study and what you are doing with it in your study or what interests you about it. This contribution should be developed from the theoretical conversations in your discipline and should reflect a specialized knowledge of your discipline. For example, the new contribution you might be making is to begin to suggest the communication processes by which political beliefs are transmitted within families. You know that such beliefs (the theoretical construct) get transmitted. Your new contribution will be to explain some of the processes by which the transmission happens. Meeting this criterion in your research question forecasts the contributions to the discipline you'll discuss in your conclusion.

The Capacity of Surprise

Your research question should have the capacity to surprise. You should not already know the answer to the research question you're asking. You want to be surprised by what you find out. If you already know the answer to your question, you don't need to do the study. Moreover, if you know the answer, you aren't really doing research. Instead, you are selecting and coding data to report on and advocate for a position you already hold.

So, for example, using the data of immigrant narratives, a research question might be, "How do traumatic events produce long-term negative effects on individuals?" This already assumes that immigration inevitably traumatizes individuals and there are no possibilities other than to experience immigration negatively. There is not likely to be any surprise in the findings because the question articulated what was expected. Continuing in this direction, one could have found examples of negative effects, but the contribution to the discipline (and future ability to publish) would have been greatly diminished.

Be Robust

A research question that is robust has the capacity to generate complex results. Your question should have the capacity to produce multiple insights about various aspects of the theoretical construct you are exploring. It should not be a question to which the answer is "yes" or "no" because such an answer is not a complex result. Research questions that typically produce robust findings often begin with:

  • What is the nature of . . .
  • What are the functions of . . .
  • What are the mechanisms by which . . .
  • What factors affect . . .
  • What strategies are used . . .
  • What are the effects of . . .
  • What is the relationship between . . .
  • Under what conditions do . . .

You undoubtedly have seen dissertations or journal articles in which there is more than one research question. Should you have more than one question in your study? Maybe, but it is discouraged. In some cases, studies contain more than one question because researchers have not thought carefully enough about what they want to find out. As a result, they take a scattershot approach and try to get close to the question they want to answer by asking about many things. A better approach is to aim for one research question and to think carefully about what it is. Refine it sufficiently so that it really gets at the key thing you want to find out.

Another reason studies sometimes include many research questions is because students confuse research questions with the questions they will use as prompts for coding their data. The many research questions are really just guides for coding data. In one study about online chat rooms and whether they have the capacity for deep culture, you may find this list of research questions:

  • What artifacts do chat rooms use as the basis for developing culture?
  • What norms characterize chat rooms?
  • What processes are used to socialize new members into chat rooms?
  • What mechanisms are used in chat rooms to repair breaches of organizational norms?

These questions are not separate research questions as much as they are questions that the researcher will use to guide an analysis of the data. They are methodological guidelines that will help in the coding of the data. Remember that a research question is what the dissertation is about. It produces the title of the dissertation.

There are some cases when more than one research question is warranted. When a study has more than one research question, it tends to be when basic information about a theoretical construct does not exist, and you need to know basic information before you can investigate a process that characterizes the construct.

Be sure to spend time making sure your research question is a good one before you get too far along in your study.