Does Living Near Classmates Help Introductory Economics Students Get Better Grades?

Abstract

This article examines whether first-year students in introductory economics courses get better grades if they have other students in their on-campus residential unit who either are taking the same course or have taken the course in the past. The study uses nine years of data for the introductory economics course at Reed College. The author finds that having dorm mates who are currently taking the class seems to have some benefit for students, but there is no evidence of benefit from having coresident students who have previously completed the class.

Keywords:

• classmates
• dorm mates
• peer effects

JEL code:

• A22

Acknowledgments

The author thanks Andreea Jurculet, who provided outstanding research assistance on this project. He also thanks Jon Rivenburg for providing the necessary data and Jim Grant, Jan Crouter, Noel Netusil, Sam Allgood, and other participants in the AEA January 2009 session on instruction for useful comments.

Notes

1. See Manski (1993).

2. Inkelas et al. (2008) provide a typology of kinds of such programs and review the literature describing them.

3. Because Reed is a small, selective college drawing its student body from a national applicant pool, the vast majority of first-year students arrive on campus knowing no other Reed students. No entering student at Reed would have a pre-established peer group of the kind that might occur when 25 graduates of the same high school attend a large state university and remain in contact with one another.

4. In addition to dorms, Reed houses higher-level students in two apartment buildings adjacent to campus. Conversations with students who lived in dorms and those who live in apartments suggest that residents in the apartment buildings do not interact in the same ways as those in dorms. The apartment buildings are larger (about 60 students) and lack the shared kitchen, bathroom, and social-room spaces of dorm units. We thus did not consider individuals who lived in the apartment buildings to be sharing a residential unit for purposes of this study. The results are not sensitive to this assumption as very few first-year students live in apartment housing.

5. The converse is not true. Reed has five language houses and, as noted above, two apartment buildings to which first-year students are rarely assigned.

6. Dummy variables for instructor were examined and proved not to be significantly associated with grade.

7. The divisions are, of course, somewhat arbitrary. Approximately 22 percent of students during the sample period earned an A or A− in Econ 201, and another 15 percent earned a B+, so these categories correspond roughly to the top quarter and the top three-eighths of students in the course. Lowering the standard to B or better would include more than the top half of students (57 percent). Other measures of quality were explored in earlier versions of the study, including the average grade of dorm mates and the best grade received in Econ 201 by a dorm mate. These measures were never statistically significant and had the added disadvantage of being undefined for the (many) students with no dorm mate who had completed Econ 201.

8. In most cases, the no-SAT students submitted ACT scores. Only the ACT composite score is retained in the college database, which does not allow us to decompose these scores into math and verbal components.

9. An alternative method, which we have explored in detail, is to use multiple stochastic imputation to impute the values of the missing cases. Multiple imputation proved problematic in this application because of observations with pervasive missing data and because imputed values for variables such as SAT and high school GPA were frequently out of range. Alternative regressions with the full 237 students using only the reader rating yielded results that are nearly identical to those reported below. Moreover, there is only a small and statistically insignificant difference between the groups with and without SAT scores in admission ratings and Econ 201 grade.

10. An alternative measure would be the proportion of dorm mates either taking or having taken the course. This would be a preferable alternative if there were large variation in the size of dorm units or if students did not become at least somewhat familiar with all dorm mates. At Reed, the size of the dorm units is sufficiently homogeneous that the proportional and absolute measures would not differ very much. Moreover, given that Reed students achieve at least some familiarity with all of their dorm mates, studying-together or mentoring opportunities depend on the number of potential partners, not the proportion.

11. This low R ² is characteristics of all grade regressions using Reed College data, not just these for the economics course. In studies predicting cumulative GPAs of college students, admission variables typically explain about 40 percent of variation; at Reed such variables explain at best 25 percent.

12. We must be careful to note that these coefficients are not true partial effects. A student with a higher SAT score (of either type) will, other things being equal, have a higher reader rating as well because higher scores will raise the admission deans’ assessment of the student. Thus, a zero coefficient on an SAT score in these regressions would mean that the weight attached to the score in the reader rating exactly matches the weight of that score in predicting the Econ 201 grade. A positive coefficient means that the score is even more important for Econ 201 success than in forming the admission office's overall assessment. A negative coefficient on one of the SAT scores is plausible; it would mean that the admission deans put more weight on that SAT component than its importance in predicting the Econ 201 grade.

13. When the number-of-classmates-in-dorm variable is treated as a set of dummies, the effect remains statistically significant with the strongest effect at three classmates/dorm mates.

14. Another study finds strong evidence that collaboration by Reed economics students takes the form of mutual contributions to homework assignments rather than free-riding by one student on the work of another. See Parker (2010).