5.6. How Do I Write a Research Proposal?

In the last chapter, we discussed what the stages in research design might look like:

1. Develop an initial research question—what is called a working research question, given that it is a work in progress (Chapter 4: Research Questions)
2. Decide on an overall research strategy—whether your analytical approach will be inductive or deductive, what the target population is, and so on. (Chapter 4: Research Questions)
3. Conduct a literature review to identify a gap in the existing body of research (a research problem) and refine your research question (Chapter 5: Research Design)
4. Decide on a method of data collection and analysis (Chapter 5: Research Design and the chapters devoted to each method)
5. Propose hypotheses based on your literature review (Chapter 5: Research Design)
6. Decide on your sampling strategy (Chapter 6: Sampling)
7. Define key concepts and measures—what is called conceptualization and operationalization, respectively (Chapter 7: Measuring the Social World)
8. Identify and address any ethical concerns about your proposed study (Chapter 8: Ethics).

As we noted, decisions about research design do not necessarily occur in sequential order. For example, you might have to think about potential ethical concerns before zeroing in on a specific research question or sample, since such considerations will limit whom or what you can study. Similarly, your method of data collection might hinge on what population of interest you want to generalize about—or vice versa. As you design your research project, you’ll be constantly moving back and forth between steps on the list below. Sometimes researchers even jump ahead to data collection and analysis, and then circle back to refine their research question. Although not ideal, this approach sometimes makes practical sense. Maybe you’re using an existing dataset and need to know what concepts are adequately operationalized within it before you decide on what to study. Or maybe you need to talk to some people before you can identify and read up on the underlying variables driving a particular relationship that your research question examines.

All that said, a good literature review sets you up nicely for choosing among possible methods of data collection and analysis. You know what sorts of research have already been done on your topic, and you have identified a compelling research problem. Now you need to decide on a methodological approach to answer your research question and address your research problem. We will learn more about the following methods in the remainder of this textbook:

1. Ethnographic observation
2. In-depth interviews and focus groups
3. Experiments
4. Surveys
5. Historical analysis
6. Content analysis
7. Social network analysis

So many different research designs exist in the social scientific literature that it would be impossible to include them all in this textbook. We cover the most popular methodological approaches in sociology, which we hope will give you a foundation for understanding more advanced designs as well. We encourage you to learn about other methods and more specialized study designs not reviewed in this textbook, which will expand your toolkit for conducting solid research.

In this section, we’ll discuss how to write a research proposal, which is a plan for the design of your study. The first consideration is to decide on a general methodological approach.

Choosing a General Methodological Approach

The rest of this textbook is devoted to describing specific research methods, so we will have much more to say about each approach and its advantages and disadvantages. At this point, we want to make some general comments about choosing between qualitative and quantitative methods. As we have noted, the distinction between qualitative and quantitative research methods boils down to the kind of data we sociologists collect and analyze. Quantitative research employs numeric data, such as scores and metrics. Qualitative research relies mostly on nonnumeric data, such as quotations from interviews and details from visual observations. As a result, qualitative research is not well-suited to be analyzed with statistical procedures. Sometimes, qualitative data is tabulated quantitatively—based on how frequently certain codes come up, for example. Many qualitative researchers reject this coding approach, however, as a futile effort to seek consensus or objectivity in a research approach that is essentially subjective.

The strength of qualitative methods, as we’ve emphasized, is in the richness of its data—the fact that stories and meanings are not collapsed into numbers but captured in all their complexity and detail. This thick description creates a sound empirical foundation for theorizing about how causal mechanisms operate and how people understand their social realities. The most frequently used qualitative research method is in-depth interviews (face-to-face, phone, or online, conducted with individuals or focus groups). A second technique is ethnographic observation. Observational techniques include bystander observation, where the researcher is a neutral and passive external observer, and participant observation, where the researcher is an active participant in the phenomenon and their input or mere presence may influence the phenomenon being studied. When qualitative researchers talk about being “in the field,” they’re talking about using one or both of these approaches to be out in the real world and involved in the everyday lives of the people they are studying. A third qualitative technique is qualitative content analysis of documents (books, emails, annual reports, financial statements, news articles, websites, etc.) or media (videos, movies, songs, etc.), which does not have to occur in the field.

In sociology, quantitative research is usually conducted using surveys. Typically, a researcher creates a quantitative dataset by asking people to answer multiple-choice survey questions and then counting their responses. (Alternatively, the researcher analyzes an existing dataset based on someone else’s survey.) Using scientific sampling procedures, sociologists can be confident that the subset of people they interviewed more or less represents the population they want to know about. That means the results of the survey—the percentage of people who had a certain political opinion, for instance—can be generalized to that larger population. Furthermore, because of the ways that quantitative measures simplify reality—boiling down wide-ranging opinions on a particular policy to several response categories, for instance—surveys allow people’s responses to be easily compared across groups, time periods, and the like. They also allow for the strength of relationships between variables to be calculated: for instance, how much a person’s gender affects how much money they make doing the same job, or how much being born in a certain part of the country affects a person’s happiness and health. Quantitative content analysis and social network analysis apply similar approaches to other forms of data (text and media for the former, and network data for the latter), tallying up phenomena in ways that can ideally say something meaningful about larger populations. (Our textbook does not currently cover social network analysis, but you can get a sense of what this methodological approach covers in the sidebar Mapping the Social World.)

While less common in sociology than in sister fields like psychology, experiments trade some generalizability for greater confidence about causality. By having control and experimental groups that are exactly the same except for the presence of a stimulus or treatment, experiments allow researchers to infer more directly that a change in one variable leads to a change in another. They can also quantify the size of that effect—in this case, with more certainty that it is causal.

For all these reasons, quantitative methods add more precision and universality to our understanding of a phenomenon than qualitative methods are capable of. At the same time, the quantification of data inevitably strips away nuance and complexity, possibly flattening meanings and oversimplifying relationships. To avoid the pitfalls of each approach, a mixed-methods research design uses qualitative and quantitative techniques jointly within a single study, with the hope that multiple methods can complement and compensate for one another.

So what sort of research design should you pursue for your study? Ideally, your research design is determined by your research question. If your research question involves, for example, testing a new policy intervention, you will likely want to use a deductive experimental design. If you want to know the lived experience of people in a public housing building, you probably want to use inductive qualitative methods like in-depth interviews and ethnographic observation. In other words, you should pursue a research design that gives you the best chance of arriving at a compelling answer to your research question. We’ll talk more about which research methods are best at answering which research questions in later chapters.

As a researcher, you have to choose a research design that not only makes sense for answering your research question but also is feasible to complete with the skills and resources you have. For one thing, the design of your research study determines what you and your research participants will do. In an experiment, for example, the researcher will introduce a stimulus or treatment to participants and measure their responses. A content analysis may not have participants at all, and the researcher may simply be reviewing organizational materials or news media to understand cultures and attitudes. Your personal preferences and talents working with data and other people will naturally lead you to pursue certain methods of research.

All research projects also require resources to accomplish. Make sure your design is one you can carry out with the time, money, and assistance available to you. For instance, are you interested in better understanding the day-to-day experiences of maximum-security prisoners? This sounds fascinating, but unless you plan to commit a crime that lands you in a maximum-security prison (generally not a wise choice even for an ambitious researcher), gaining access to that facility would be difficult. Perhaps your interest is in the inner workings of toddler peer groups. If you’re much older than four or five, however, it might be tough for you to access that sort of group. Your ideal research topic might require you to live on a chartered sailboat in the Bahamas for a few years, but unless you have unlimited funding, it will be difficult to make even that happen. While the topics about which research questions can be asked may seem limitless, there are limits to which aspects of topics we can study and how we can study them.

One of the most important questions in feasibility is whether or not you have access to the people you want to study. For example, let’s say you wanted to better understand students who engaged in self-harm behaviors in middle school. That is a topic of social importance, to be sure. But if you were a principal in charge of a middle school, would you want the parents to hear in the news about students engaging in self-harm at your school? Building a working relationship with the principal and the school administration will be a complicated task, but it will be necessary to gain access to the population you need to study. As we discussed in Chapter 2: Using Sociology in Everyday Life, research must often satisfy multiple stakeholders—that is, individuals or groups who have an interest in the outcome of the study. Your goal of answering your research question can be realized only when you account for the goals of the other stakeholders. School administrators also want to help their students who are struggling with self-harm, so they may support your research project. But they may also need to avoid scandal and panic, providing support to students without making the problem worse.

Assuming you can gain approval to conduct research with the population that most interests you, do you know if that population will let you in? Researchers like Barrie Thorne (1993), who study the behaviors of children, sometimes face this dilemma. In the course of her work, Thorne has studied how children teach each other gender norms. She also studied how adults “gender” children, but we’ll focus on just the former aspect of her work. Thorne had to figure out how to study the interactions of elementary school children when they probably would not accept her as one of their own. They were also unlikely to be able to read and complete a written questionnaire. Since she could not join them or ask them to read and write on a written questionnaire, Thorne’s solution was to watch the children. Although this seems like a reasonable solution to the problem of not being able to actually enroll in elementary school herself, there is always the possibility that Thorne’s observations differed from what they might have been had she been able to actually join a class. What this means is that a researcher’s identity, in this case Thorne’s age, might limit (or enhance) her ability to study a topic in the way that she most wishes to study it.

In addition to personal characteristics, there are also the very practical matters of time and money that shape what you are able to study or how you are able to study it. In terms of time, your personal time frame for conducting research may be the semesters during which you are taking your methods courses or working on a thesis. Future employers will give you specific deadlines for completing research tasks. Those timelines will shape the sort of research you are able to conduct. Surveys can be completed in minutes online, but recruiting a representative sample for that survey might entail much more work. Immersive qualitative work—such as embedding yourself in an organization—can take months or years. Money, as always, is also relevant. Obtaining commercial datasets can be expensive, and finding participants willing to sit for long in-depth interviews often requires you to pay them.

Writing the Sections of a Research Proposal

Once you have decided on a general methodological approach that is appropriate and feasible, you will want to start describing your planned study in a research proposal. Think of your research proposal as a literature review plus sections devoted to methods, hypotheses, and limitations. We have already described what goes into the literature review, but note that your review of past studies also allows you to generate hypotheses regarding what results your research is likely to uncover. The proposal’s methods section, in turn, will describe in depth the procedures you will use to answer your question. A proposal often concludes with a discussion of possible implications and limitations of the proposed research. In other words, the research proposal is basically a draft version of your final empirical paper—minus the all-important results section, and including a preliminary discussion of implications that you are certain to revise in the final paper. We will cover each of these sections in your proposal.

Introduction. As we noted earlier, the introduction to your research proposal looks the same as an introduction for a stand-alone literature review. However, in both the abstract and the parts of introduction that summarize the proposed study, you will have more details to provide. We recommend the following structure for this summary:

Past studies have failed to examine [research problem]. Using [methods/data], this proposed study examines [research question]. The findings of this study are expected to be [hypotheses], meaning that we need to rethink [implications].

Literature review. The literature review follows the proposal’s introduction, and looks exactly as we described in Section 5.1.

Methods. The methods section is a plan for what exactly you will do in your study. The more specific it is, the better. Of course, at this planning stage, many of the decisions you make will be somewhat arbitrary. For instance, you may not know exactly how many people you will be able to recruit for the in-depth interviews or surveys you want to do. It may be hard even to guess how large a sample is reasonable. Nevertheless, put down some numbers, however preliminary. Those details can always change as you talk through the project with other people and figure out what is feasible.

Let’s break down what goes into the methods section of your proposal. (Although some of the terms we mention might be unfamiliar to you at this stage, when you return to these guidelines after reading the following chapters, they will be much clearer!) Overall, this section of the proposal should provide a thorough discussion of the methods, data, and measures you will use to test your hypotheses, generate theory, and otherwise conduct your research. Justify your choice of methodological techniques, sampling procedures, interview and survey questions, and so on, referring to any strengths and weaknesses in the methods you plan to use, given what your research problem and research question are. There are always shortcomings or compromises involved with any proposed sampling procedure, operationalization of a theoretical concept, or other methodological choice, so you should have plenty of things to discuss in this section.

For all studies, you should discuss the proposed target population, sample size, time frame of data collection, sampling techniques, and process for recruiting any respondents. You should also review the key measures that you are focusing on, describing exactly how the underlying concepts of interest were or will be operationalized in terms of actual question wording and other specifics. Finally, discuss your analysis strategy (e.g., what statistical procedures will be used, or how qualitative data will be coded). For a proposed survey study, make sure to provide substantial detail about how the survey was or will be fielded, discussing how representative the sample is of the population of interest (including any weighting or other adjustments) and identifying the dataset and sponsoring organization (if using a secondary dataset). In a subsection called “Measures,” discuss the department variables and then independent variables.

For a proposed qualitative study, include a subsection called “Case Selection” in which you elaborate on the purposive sampling strategy used in choosing sites and types of respondents during data collection. Discuss the specific inclusion and exclusion criteria for your sample, and be specific about the process for recruiting interviewees or gaining access to sites. For studies based on in-depth interviews or surveys, you should create a full interview guide or survey questionnaire to accompany your proposal, in addition to describing your key variables in the methods section or measures subsection in as much detail as possible.

Hypotheses. For quantitative studies in particular, you will want to include hypotheses—likely answers to your research question—in your proposal, which typically follow the methods section (but can also appear in the literature review, usually at its end). For qualitative studies, sometimes the hypotheses are merely implied, and if they are mentioned, they may not be identified as “hypotheses” per se—just listed as expectations in the literature review or elsewhere.

A study’s hypotheses should be derived from the scientific literature or based on logic. After you decide on the study’s overall methodology and conduct a literature review, you should be able to apply what you know from the literature to the sample or context you’re studying, specifying what results you expect your research to uncover.

As scientists, we test hypotheses using data. This process is called inference. We infer—conclude based on evidence—whether or not a hypothesis is valid. When the data don’t support our hypothesis, it is rejected—or falsified. As we discussed in earlier chapters, your hypotheses must be falsifiable—that is, you must be able to confirm or disconfirm them with data obtained from your target population.

Hypotheses can be strong—in that they propose a specific and causal relationship—or they can be weak—in that they just say the two variables are (somehow) associated. Causal inference—testing whether changes in one variable truly cause changes in another variable—is a heavier lift than just testing whether the two variables are correlated. Consider these different hypotheses:

Weak: Children’s family incomes are related to their later educational attainment. (This tells us nothing about the direction of the hypothesis—whether the relationship is positive or negative—or its causality—whether greater family income actually causes higher educational attainment.)

Stronger: Children’s family incomes are positively related to their later educational attainment. (This hypothesis indicates the directionality but not the causality of the relationship.)

Strongest: Children’s family incomes have positive effects on their later educational attainment. (This hypothesis specifies both the directionality and the causality—that greater family income causes higher educational attainment.)

Scientists typically want to test strong hypotheses because they want to know the directionality and causality of a particular relationship. That said, sometimes the data available won’t allow you to test those aspects of a relationship; as we’ll discuss in Chapter 12: Experiments, certain research designs are better at evaluating whether a relationship is truly causal. Regardless of whether our hypotheses are weak or strong, however, they should clearly specify independent and dependent variables. In the hypothesis, “children’s family incomes have positive effects on their later educational attainment,” it is clear that income is the independent variable (the “cause”) and educational attainment is the dependent variable (the “effect”). It is also clear that this hypothesis can be evaluated as either true (if higher family income raises later educational attainment) or false (if higher family income has no effect on, or lowers, later educational attainment).

Note that hypotheses can be simply described within the literature review or provided in a separate section. Sometimes, they are listed formally as hypothesis 1, hypothesis 2, and so on. You should follow the lead of similar studies found in your literature review, given that different fields and subfields have different norms for presenting hypotheses.

Conclusion and limitations. There isn’t much to say in the conclusion to your research proposal, since you haven’t yet conducted your study. Thus, this section will necessarily be short. That said, you should be able to broadly discuss the possible implications of your study for research, theory, practice, or policy, giving particular attention to how your study connects to the existing literature. You should circle back to the research problem that you argued for at the end of your literature review or in a separate “Statement of the Problem” section, making the case for how your study will resolve that problem. If the limitations of your study were not fully discussed in the methods section, you should point them out, given the relevant methodological and theoretical approach.