CLASSES AS RESEARCH TEAMS

For good reason, professors may balk at the idea of incorporating the class into their own research. Yet, with the following precautions in place, this can be advantageous for both the professor and the students. First, a professor who works with a class on his or her own research needs to be explicit. Second, under no circumstance should the professor view or treat the class as a team of research assistants. To this end, the students need to be treated as at least partial partners in the project, offering them some control over aspects of the project. Finally, if the students contribute substantially, professors need to entertain the idea of making them coauthors—while this may lead to an enormous list of coauthors, it would not be unusual relative to publications in medicine and some sciences.

The first example comes from a course on media and campaigns. An overarching question for such a class is: do campaigns matter? I devote the first few weeks of the course to discussions about how campaigns may matter—particularly focusing on how the media may affect opinions. I then raise the question to the class: how can we test the proposition of whether media campaigns matter? The students invariably conclude that this can best be tested with a survey on campaign day—an exit poll. Here, research and teaching come together. We spend considerable time discussing (1) how such a poll can be done from a methodological point of view (e.g., how do we choose a sample and measures?; questions that can be discussed after a brief primer sans prior knowledge of statistics), (2) how it can be used to assess distinct types of campaign effects, and (3) how we should implement the survey. I explain to the class assessing the impact of campaigns is a pressing research question, and we can address it. I also fully disclose that, while they raised the questions, the readings and discussion was meant to steer them, although not with certainty, to such a study. Then, I ask the class if they would like to work on such a study, and in each of the three instances of my teaching this class, the students unanimously and enthusiastically affirmed their desire to do so.

Next, we work together to create a study design that involves five steps: (1) determining a content analysis approach to capture campaign communications, (2) identifying the ideal polling places to implement the survey (via random selection), (3) completing the Institutional Review Board process and learning about the state laws for polling and contacting the relevant electoral officials, (4) designing the survey, and (5) isolating the practical steps of implementation (e.g., designing badges to identify pollsters and obtaining supplies, such as clipboards and pens). In 2000, with my guidance, my class realized a unique opportunity given they lived in Minneapolis-St. Paul which, at the time, was one of the few remaining cities with two major newspapers. Thus, the students divided themselves into two teams with one team content analyzing the Minneapolis Star Tribune and the other analyzing the St. Paul Pioneer Press. While I advised them in developing the content analysis instrument (e.g., how to specifically code for issues and image), I was conscious to enfranchise the students by merging my ideas with theirs. This also was how we went about developing the survey itself, which included extended classroom discussions about how to most accurately gauge media exposure and attention. In the end, when we linked the content analyses to exit poll respondents’ self-reported newspaper reading habits and their opinions, we found that the distinct newspapers dramatically influenced opinions; controlling for a host of other variables, those exposed to the Star Tribune expressed evaluations based on the information in that paper (which lead to a more positive evaluation of one candidate) while readers of the Pioneer Press displayed distinct assessments consistent with what that paper reported. The evidence starkly revealed that news exposure shapes candidate evaluations.

After the students and I finished analyzing the data, we discussed what happens next. I detailed the publication process and asked whether the class felt it reasonable that I proceed and attempt to publish papers using our research. The students unanimously supported this, and indeed the work resulted in four publications; one of these publications revolved around a question that the students themselves pressed to include on the survey (i.e., time of voting decision). While the benefits to me are clear, it is important to accentuate the students’ reaction to the process. As discussed, I went to great lengths to ensure a transparency and open discussion about what product may result and authorship issues (along these lines, early in the class we discussed what authorship entails and worked through various hypothetical scenarios). I also asked the students to anonymously complete a brief response on their feelings about authorship, and none felt they were entitled to be coauthors as they all viewed it as a learning classroom exercise that I had overseen. Had even a single student felt uncomfortable with the authorship plan, I would have remained open to including a subset or all students (it is important to err on the side of inclusion). In terms of student reactions, their course evaluations suggested the project to be a highlight of their quarter, as most commented that they learned how to address a social science problem. Indeed, half of the class went on to complete senior theses in the field of political behavior. Even those less enthralled with being part of the research process enjoyed the exit polling experience, about which I asked them to write a brief paper. Most expressed fascination with talking to people as they were voting and learning what was on their minds. In short, all indications, including continued contact with the students over future years, suggest it was a successful pedagogical exercise. I replicated it again twice, although in each of those cases the class independently decided to focus on distinct campaign dynamics (e.g., questions of identity or attitudes about salient scientific issues discussed in the campaign) and that shaped the project we undertook. Allowing students to play a major role in determining the direction of the research gave them a sense of ownership and satisfaction in seeing the work come to fruition.

My other two examples differ insofar as rather than working with the class, from the start, to conceive, develop, and carry out a research project, I began the class by presenting a broader project to which they would add. Specifically, I developed these two courses with the explicit intent of aggregating student projects to contribute to larger data collection efforts. As with the first example, I was explicit about this goal and designed the students’ work in a way that the individual contributions themselves constituted complete research projects.

One of these classes focused on how the media (e.g., newspapers) covers issues. I explained to the class (also detailed on the course description that was available prior to student registration) that my goal was to collect content analyses of news coverage (newspapers) on a large range of issues. We spent the first several classes discussing theories of news coverage including discussions of liberal bias, balance, gratification theory, negativity bias, episodic versus thematic coverage, and over-time trends in coverage. Next, we moved into the mechanics of how to assess these theories by focusing on “frames in communication,” which are points of emphasis in the discussion of a particular issue. At this point in the course, each student chose a particular issue from a large number of options on which he or she would focus (e.g., the Patriot Act, global warming, Social Security, same-sex marriage, Bush v. Gore, immigration, death penalty, and hate group rallies). The student’s individual project then was to evaluate news framing on the given issue.

I asked each student, after having read a few examples of prior frame content analyses, to develop a plan for how he or she would complete the assignment (that would then be shared with all members of the class). This proved to be a critical activity as the range of methods proposed equaled the number of students in the class. We then worked together as a class to make methodological choices including which frames for each issue to identify (e.g., based on analyses of interest group communications, Supreme Court decisions, and other news coverage), time periods to cover, sources to code (e.g., which newspaper), key words to identify articles, features of the frame to code (e.g., negativity, episodic, negation), and how to assess reliability. The students ostensibly marveled at the host of challenges faced but then effectively worked to arrive at a cohesive approach that each tailored to his or her particular issue. The result was approximately 15 research papers on individual issues. Not only did the students, without exception, express amazement at their abilities to conduct such nontrivial research during a quarter, but several of the projects uncovered fascinating dynamics such as how same-sex marriage is framed as a tolerance issue to a much greater extent in Canada compared to the United States. Another author identified fascinating shifts in frames over the course of events surrounding the neo-Nazi march that had been planned, but failed to occur, in Skokie, Illinois, in 1978; the author of this paper even went so far as to find and interview the newspaper reporter who covered much of this event for the Chicago Tribune. Toward the end of the quarter, I combined all of the results, and, as a class, we discussed the implications of the coding for my larger project that demonstrated trends of a negativity bias, a liberal bias, over-time changes with a tendency toward a reduction in the number of frames used, and the preference for episodic as opposed to thematic coverage. Then, I merged those data with a host of additional content analyses (some on the same issues and some on others) to produce what, to the best of my knowledge, constituted the most exhaustive frame content analysis to date, and resulted in a published paper. Most importantly, however, I believe the project enabled students to complete a challenging research project and experience the (near) culmination of a larger project in which they played a critical role. It is telling that, again, more than half the students continued their projects as parts of senior theses.

A third similar example, of which I will not go into great details because the pedagogical elements align with the just-described example, involves students in a class coding congressional candidates’ websites. Every congressional election year, I teach a course on campaigns. We spend the first part of the class learning about and critiquing extant theories of campaigns including issue ownership, going negative, image management, and technological campaign options. Next I raise this question: how can we test these theories? Students offer a number of suggestions (e.g., television advertisements, public addresses, newspaper coverage) but, without fail, they confront a sampling challenge of how one can draw a representative sample from the population of House and Senate campaigns. The answer we invariably arrive at involves exploring candidate websites because virtually all candidates have had such sites since 2002. Students then work with me to update a coding document that enables testing the theories of campaigns we discuss (i.e., we reviewed theories and hypotheses and the students insert themselves in the testing part of the project). The document is “updated’ because it is a living coding analyses framework that began in 2002. When we agree on a final coding framework (during which time we discuss issues of measurement and validity), students code about six websites and write research papers that (1) discuss what they find in light of the aforementioned theories, and (2) offer suggestions on how the sites could be improved. Then, we bring together all the web data to test the theories with a larger sample and explore whether those who followed the prescriptions of different theories fared better, all else constant. As in the prior example, students finish the quarter with an impressive research paper, the satisfaction of contributing to a larger project, and in many cases a basis for what becomes a senior thesis.

In each of these three examples (and I could offer several more), students learn the entire “idealized” scientific method—whether through involvement at each step as in the exit poll or involvement in the data collection as in the latter two examples—and learn of the challenges and imperfections. In all cases, I am clear that the work will likely end up as a part of a publication, which makes it all the more important that the students not be treated as assistants helping with my research. Rather, they are given ownership either as a group or on individual projects and in the end produce their own serious research that often leads to further work. The universal feedback suggests feelings of exhilaration when the student realizes that in just nine or ten weeks (the length of a quarter), he or she produced a substantial research project that generated knowledge to which, for some time, only he or she possessed access. In the prototypical case, the resulting project is a solid piece of research that sometimes is extended into a senior thesis (and, in theory, this work could turn into an independent publication by the student). Moreover, at times, the student inquires about opportunities to work collaboratively with me, in which case coauthorship relationships (with me) evolve. The point is that teaching and research are not treated as distinct endeavors but are rather merged in such a way that learning how to do research becomes the goal of teaching and that in turn results in the production of new research.

CLASSES DOING RESEARCH PROJECTS

In many cases, working with a class as a research team is not feasible given the nature of the topic and/or the size of the class. These situations, however, present an alternative option where the students do not work toward a broader aggregated goal but instead independently, step-by-step, do research with the potential of later expansion and publication. I offer two such examples.

The first comes from a general undergraduate (junior/senior) seminar on public opinion. In this class, the central assignment for students is to produce a significant public opinion research paper. The first class begins with me informing the students that, by the end of the quarter, they will produce a 20- to 25-page research paper. The students respond with looks of confusion and trepidation. Then, I explain how we will do this step-by-step such that by the end of the quarter, their job will largely involve revising parts and putting pieces together to arrive at such a substantial piece of research. We then go week-by-week through each step of the research production process: identifying a question, conducting a literature review and finding a gap, proposing a theory (even if rough and intuitive) and generating hypotheses, devising a research design (in this case involving either a survey or an experiment), developing an approach for sampling and measurement, collecting their data, analyzing it, and then describing the results in light of expectations. Students turn in each of these pieces on a weekly basis, receive feedback and revise accordingly such that by the end, they turn in a significant piece of novel research (which often does not require or need sophisticated statistical analyses and thus there is no prerequisite for the class). The initial looks of trepidation evolve to feelings of great accomplishment and even joy at the fact that the student at some point obtains an insight that is uniquely his or hers. In a number of cases, students continue working on these projects, often asking me to join them (as in the aforementioned case where students do their own research), and the result, thus far has been a number of published papers in the American Political Science Review, the American Journal of Political Science, and the Journal of Politics. Thus, my own research agenda inadvertently benefits in large part through the training of students in how to conduct research which in turn stimulates their desire to do more. As with the prior examples, the blending of research and teaching mutually benefits me as the professor and the students who in many cases have gone on to use what they learned in a wide range of professions.

A final example comes from a course where carrying out a full-fledged research project did not fit with the other dictates of the class. Yet this did not preclude me from having the students address a tricky research question. Specifically, in a class on mathematical political science, we discussed the many challenges of measurement. We reviewed a host of techniques for addressing the measurement of sensitive behaviors such as drug use among college students. The measurement techniques included assurance of confidentiality, normalization (e.g., telling respondents others engage in the behavior so it’s “normal”), self-administration sans face-to-face contact (e.g., on a computer), sealed booklets, and randomized response techniques (or list experiments) (e.g., Tourangeau and Smith Reference Tourangeau and Smith1996). Then, I asked the class to develop an approach for comparing the techniques. They worked together and devised a clever experimental approach such that they identified a sample of students and then randomly assigned them to report illicit drug use using each of the aforementioned approaches. The critical insight at which the class arrived is that the list experiment (where respondents report the number of behaviors in which they have engaged from a list of three or four—they are randomly assigned to the list length) ostensibly provides the most accurate aggregate percentage of drug usage (i.e., by comparing the disparity between the 3-item and 4-item list). Yet, the aggregate results from list experiments prevent determination of whether a given individual has used drugs, which is often desirable if we hope to causally explain drug use. The class found that the sealed booklet, by far, offered the most accurate estimate as it strongly resembled, in aggregate percentages, the list experiment. Not only is this a novel insight—that a sealed booklet rather than, for example, self-administration is the most valid individual level measurement approach—but it also taught the students how to address a relevant question in a systematic manner. In the end, the palpable student satisfaction with the experience highlighted how motivating research projects can be for students who, again, in this case often went on to write senior theses that revolved around different measurement issues.