So you want to go to graduate school? Factors that influence admissions to economics PhD programs

Abstract

The authors survey admissions coordinators about the importance of application components in admissions decisions for economics PhD programs. The survey explores the importance of difficult-to-quantify aspects such as a targeted personal statement, strength of letters of recommendation, extracurricular activities, and related work experience. The most important aspects of an application are GPAs in math and economics, letters of recommendation, and GRE quantitative score. The strength of letters of recommendation carries more weight than the prominence of the letter writer. Top-25 programs place a higher value on undergraduate program rank (for students from both domestic and international universities) and strength of letters of recommendation.

Keywords:

• Economics PhD
• graduate admissions
• undergraduate education

JEL CODES:

• A11
• A22
• A23
• I21

Acknowledgments

The authors thank KimMarie McGoldrick for the list of economics PhD programs and helpful comments on the article. The authors also thank Gail Hoyt for organizing and chairing the AEA session centered on our paper and the panelists for that session: Gautam Gowrisankaran, Navin Kartik, Martin Boileau, Wojciech Olszewski, Daniele Paserman, and Marcus Berliant. Finally, the authors thank the participants at UNCW’s 2018 Economic Teaching Workshop, SEA 2018, and AEA 2019.

Notes

1 Becker, Rouse, and Chen (2016) focus on a narrower aspect of admission and evaluate the impact of the American Economic Association’s (AEA) summer program for minorities. They find that participants were more than 45 percentage points more likely to apply and 40 percentage points more likely to attend a PhD program in economics.

2 See Conley and Önder (2014), Grove and Wu (2007), Grove, Dutkowsky, and Grodner (2007), Schlauch and Startz (2018), Siegfried and Stock (2004), Stock and Hansen (2004), Stock and Siegfried (2006, 2014, 2015), Stock, Finegan, and Siegfried (2009a, 2009b), Stock, Siegfried, and Finegan (2011), van Ours and Ridder (2003), among others.

3 See Online Appendix A (Supporting Information) for the full survey questions.

4 Note that we asked admission coordinators to self-describe their PhD program rank rather than their institutional rank (see Q4.11 in Online Appendix A [Supporting Information]).

5 Although, a 5-point rating scale was used to collect survey data, the 5-point scale was collapsed to a less granular 3-point scale during the analysis. Results were qualitatively similar, suggesting imperceptible differences between "slightly important" and "moderately important" and between "very important" and "extremely important." Categories were combined to reduce the number of empty cells occurring when no program indicated a characteristic was extremely, very, moderately, slightly, or not important. Additionally, the combined Likert scale increased power for statistical tests of proportions for subsamples of programs.

6 We also conducted the proportions tests comparing T50 and 50+ schools. The results are largely similar. The following important, statistically significant differences emerge: when comparing T50 and 50+ schools, (1) U.S. citizenship is less important to T50 schools, (2) rank of graduate program attended by an international student is more important to T50 schools, (3) more T50 schools rate verbal GRE score as moderately important, and (4) fewer T50 schools rated SAS and R as very important programming languages.

7 Questions had an average response rate of 79 percent. Ninety-five percent of questions had a response rate higher than 50 percent.

8 Online Appendix B Table 2 (Supporting Information) contains proportions tests comparing the percentage of T25 and 50+ programs that said a characteristic was not important or moderately important. T25 programs are less likely to rank gender as not important and more likely to rank it as moderately important in admissions decisions.

9 The survey question asked respondents only about the importance of a student having taken the course and not how performance in the course factored into the admission decision.

10 p values for all differences are presented in the Online Appendix (Supporting Information). The Appendix also includes a comparison of the proportion of T25 and 50+ programs stating that a characteristic was not important. T25 schools were less likely to say that prominence of the letter writer, being a tutor, and programming skills were not important.