And the Data Say…

Hi, and thanks for reading! Chad has graciously invited me to share some activities with y’all focused on bringing research methods ideas into the substantive classroom. The goal of these activities is twofold. First, programs are increasingly adopting objectives (and even school-wide QEPs) about research literacy and data or numeric literacy. The activities I’ll be sharing support those goals.

Second, students have the best chance of success in the research methods classroom when Introduction to Research Methods is not actually their introduction to research methods. One of the reasons students have anxiety and less-than-optimal experiences in their Research Methods courses is the sheer quantity of new material. If we can decrease the novelty by even a small amount, we can improve their experiences in that often-required Methods class. Through these activities, I hope to show you some ways to incorporate social scientific thinking and methods into even introductory courses in ways that require relatively low amounts of additional preparation or even class time.

Students often ask questions about general trends or patterns. Do they usually do that? Is that what normally happens? With a little nudging, you can turn this curiosity into a data-based activity. When you reach a topic where data is readily available, consider a class activity that asks students to hypothesize about relationships between variables or indicators, then test those hypotheses on the spot using a quick graph. For example, my Intro to American Politics textbook spent a bit of time showing graphs of how partisanship has varied over time in Congress and how the current period of hyperpartisanship is different, and how partisanship affects the ability of Congress to get things done.

I parlayed this into a data activity by asking small groups of students to choose a measure of each variable from a set that I had prepared, then hypothesize about the relationship they expected to observe between those two indicators. For example, as the percentage of party-line votes in a house of Congress goes up, the number of bills passed should go down, or unified government should produce fewer bills proposed than divided government. (One group went as far as to propose that the difference between proposals and passed bills should be greatest when government is divided.) Their next task was to sketch a graph of what they expect to find. The key task here was to choose an appropriate graph form: a line graph for the first hypothesis here and a bar graph for the second hypothesis. (Our textbook, Barbour and Wright’s Keeping the Republic 8th ed. Essentials, had a brief section on data presentation that they read as preparation for this activity.)

Finally, I used classroom computer to generate a quick graph to evaluate the hypothesis. This requires better-than-average familiarity with MS Excel’s graphing features, which can be finicky to the inexperienced, or the ability to use basic statistical software (and have it installed on the classroom computer). I preferred the stats software because its ability to handle missing data is far superior, and the graphic interface is far more intuitive and manipulable, but your mileage may vary. Discussion of each hypothesis included introduction of terms like positive and negative relationships and categorical vs continuous variables, but as far as I was concerned, those were bonus.

The best thing about this activity is that you can adapt the idea to any substantive class. Good questions to explore in introductory classes might include:
• How do demographics map onto vote choice and partisan affiliation?
• How does polarization affect legislative output?
• Are democratic countries wealthier or better off than non-democratic countries?
• Do US states differ in their policies, preferences, or demographics?
• What characteristics of states or wars make alliances (un)reliable?
At the upper division level, you can get much more specific with the questions and link them more closely to research in the field. My International Political Economy class explored Paul Collier’s four ‘traps’ that keep countries of the Bottom Billion from developing economically.
Questions that involve concepts with multiple measurement options – either multiple measures, such as democracy (POLITY, dichotomous, W/S, Freedom House), or various facets of the same concept (like development as GDP per capita, birth rates, HDI, etc., or polarization as party line votes, extremeness of NOMINATE scores, etc. – tend to work best. Student pairs have more things to select from, which results in different graphs.

Where can you get readily available data? My data on partisanship and legislative output came from the joint Brookings-AEI project on the topic; I downloaded Excel spreadsheets directly from their website (tiny text at bottom). Depending on your needs, the prepared datasets available with many quantitative methods textbooks can be useful as well; download the files directly from the publishers’ websites for free (registration is often not required on the student sites). Phillip Pollock’s textbook comes with datasets on countries, suitable for many topics in comparative and international politics, and selected questions from the NES and GSS. Theresa Marchant-Shapiro’s book comes with a dataset on US states as well as a countries one with different variables. Occasionally, replication data from articles can work as well; Ashley Leeds’ alliance reliability data is fun to play with for international relations classes. For the partisanship activity, I spent about 30 minutes hunting for, downloading, formatting, and familiarizing myself with the data, and about another 10 minutes writing up a handout with instructions. That’s definitely less time than I would need to write 30 minutes of lecture for a new-prep course.

What topics in your classes would lend themselves well to data exploration? Share your ideas here!