1. 1 The Correlates of Prestige Across Graduate and Professional Schools Kyle Sweitzer Data Resource Analyst Michigan State University Fred Volkwein Professor and Senior Scientist Center for the Study of Higher Education Penn State University Paper presented at the 48 th Annual Forum of the Association for Institutional Research Seattle, WA May 27, 2008 ©2008 Kyle V. Sweitzer

2. 2 Why study reputation ratings? Prospective graduate students use graduate program ratings to inform their application and admissions decisions. Administrators use graduate program ratings to inform resource allocation decisions. (Ehrenberg and Hurst, 1996)

3. 3 Existing studies on rankings/ratings Most of the studies have examined institutions’ graduate programs as a whole, via aggregating individual program ratings (Volkwein, 1986; Grunig, 1997). Few studies have examined graduate program ratings at the department or school level. U.S. News Even fewer have looked at the U.S. News graduate school ratings (most have examined the NRC ratings).

4. 4 Research Questions U.S. News What variables relate to the U.S. News peer assessment ratings of graduate programs in the professional school disciplines of business, education, engineering, law, and medicine? Are there variables relating to prestige that are common across all of the disciplines in the study, and are there variables that are specific to certain disciplines?

5. 5 Conceptual Framework See paper

6. 6 Population America’s Best Graduate Schools Schools/Colleges appearing in the lists of “The Top Schools” in business, education, engineering, law, and medicine in the 2008 edition of U.S. News’ America’s Best Graduate Schools. 50 Schools of Business 52 Schools of Education 51 Schools of Engineering 104 Schools of Law 51 Schools of Medicine

7. 7 Variables / Data Sources DEPENDENT VARIABLE – Peer assessment survey of deans, faculty, program directors U.S. News INDEPENDENT VARIABLES – Data from U.S. News --standardized admissions tests --program acceptance rates --full-time graduate enrollment in the school --non-resident tuition --student/faculty ratio --undergraduate GPA --variables specific to a discipline

8. 8 Variables / Data Sources Research activity is measured in terms of faculty publications per capita. Institute for Scientific Information Web of Science Science and Social Science Citation Indices Search on “Subject Category” for journals specific to a discipline.

9. 9 Descriptive Statistics, Schools of Business Variable NameMeanStd Dev Peer assessment score 3.71 0.54 Average undergraduate GPA 3.37 0.10 Average GMAT 666 27.0 Acceptance rate for school 0.40 0.13 Avg starting salary of grads93,16013,167 Pct grads employed at graduation0.74 0.08 Non-resident tuition32,692 6,338 FT graduate enrollment in school449 369 FT faculty in school156 65.6 S-F ratio for school2.95 1.71 Total publications 2001-05454 239 Pubs / full-time faculty 2001-05 3.15 1.51

10. 10 Descriptive Statistics, Schools of Education Variable NameMeanStd Dev Peer assessment score3.68 0.41 Average GRE1151 86.2 Doctoral acceptance rate for school0.34 0.15 Doctoral degrees granted 2005-0658.4 44.6 Pct students in doctoral program0.41 0.13 Research expendit’s 2006 (millions)15.37 7.75 Rsch exp / FT fac 2006 (thousands)236.0 129.6 Non-resident tuition21,732 5,772 FT graduate enrollment in school 516 338 FT faculty in school75 40.4 S-F ratio for school7.74 5.21 Total publications 2001-05155 78.6 Pubs / full-time faculty 2001-05 2.53 1.41

11. 11 Descriptive Statistics, Schools of Engineering Variable NameMeanStd Dev Peer assessment score3.72 0.54 Average quantitative GRE760 11.6 Acceptance rate for school0.28 0.11 Pct faculty in Natl Academy of Eng0.06 0.04 Doctoral degrees granted 2005-0699.9 66.1 Research expendit’s 2006 (millions)88.77 53.31 Rsch exp / FT fac 2006 (thousands)482.5 186.5 Non-resident tuition24,309 6,593 FT graduate enrollment in school1380 855 FT faculty in school294 165 S-F ratio for school3.77 0.88 Total publications 2001-051384 782 Pubs / full-time faculty 2001-05 5.16 2.61

12. 12 Descriptive Statistics, Schools of Law Variable NameMeanStd Dev Peer assessment score3.09 0.77 Median undergraduate GPA3.52 0.14 Median LSAT162 4.20 Acceptance rate for school0.24 0.07 Bar passage rate0.86 0.08 Pct grads employed at graduation0.79 0.13 Non-resident tuition29,005 6,032 FT enrollment for school722 281 FT faculty for school51 21.5 S-F ratio for school14.29 2.68 Total publications 2001-0559.9 65.3 Pubs / full-time faculty 2001-05 1.02 0.83

13. 13 Descriptive Statistics, Schools of Medicine Variable NameMeanStd Dev Peer assessment score3.73 0.56 Average undergraduate GPA3.70 0.07 Average MCAT 10.86 0.57 Acceptance rate for school0.07 0.03 NIH rsch expendit’s 2006 (millions)241.8 176.1 Rsch exp / FT fac 2006 (thousands)163.2 65.3 Non-resident tuition37,332 6,683 Total enrollment for school 581 177 FT faculty for school1486 985 Faculty-student ratio2.79 2.07 Total publications 2001-057244 4159 Pubs / full-time faculty 2001-05 5.27 2.39

14. 14 Regression Analysis Using the conceptual framework as a guide, we used the peer assessment score as the dependent variable and estimated a blocked (set-wise) regression model for each of the five separate graduate/professional school disciplines. In the first block, we entered the institutional characteristics, such as the size and wealth of the school. In the second block, we entered the faculty and student indicators. In the third block we entered the variables reflecting faculty and student outcomes. We avoided collinearity by picking the strongest indicator from each set of variables.

15. Regression Results, Schools of Business Standardized Betas of Significant Coefficients VariablesModel 1Model 2Model 3 Full-time enrollment.624***.407***.267* Non-resident tuition.330***.228* Student-faculty ratio ns Avg GMAT score.388***.253** Pubs per faculty 2001-2005 ns Starting salary of grads.596*** Adjusted R-Square.736.807.878 ---------------------------------------------------------------------------------------------------------------- *Significant at.05 level; **Significant at.01 level; ***Significant at.001 level. ns =non-significant when entered into model 15

16. Regression Results, Schools of Education Standardized Betas of Significant Coefficients VariablesModel 1Model 2Model 3 Full-time enrollment.366**.354*.535** Non-resident tuition.514*** Student-faculty ratio ns Avg GRE score ns Pubs per faculty 2001-2005.421* Adjusted R-Square.368.377.474 ---------------------------------------------------------------------------------------------------------------- *Significant at.05 level; **Significant at.01 level; ***Significant at.001 level. ns =non-significant when entered into model 16

17. Regression Results, Schools of Engineering Standardized Betas of Significant Coefficients VariablesModel 1Model 2Model 3 Full-time enrollment.664***.576***.792*** Non-resident tuition.327** Student-faculty ratio ns Avg GRE score.443***.226* Pubs per faculty 2001-2005.468*** Adjusted R-Square.447.619.721 ----------------------------------------------------------------------------------------------------------------- *Significant at.05 level; ***Significant at.001 level. ns =non-significant when entered into model 17

18. Regression Results, Schools of Law Standardized Betas of Significant Coefficients VariablesModel 1Model 2Model 3 Full-time enrollment.213*.159*.163** Non-resident tuition.508*** Student-faculty ratio –.207*** –.174*** Median LSAT score.712***.530*** Pubs per faculty 2001-2005.264*** Employment rate at graduation ns Adjusted R-Square.397.795.849 ----------------------------------------------------------------------------------------------------------------- *Significant at.05 level; **Significant at.01 level; ***Significant at.001 level. ns =non-significant when entered into model 18

19. Regression Results, Schools of Medicine Standardized Betas of Significant Coefficients VariablesModel 1Model 2Model 3 Full-time enrollment ns.224*.342*** Non-resident tuition ns Faculty-student ratio ns.313** Avg MCAT score.701***.637*** Pubs per faculty 2001-2005.374*** Adjusted R-Square.016.540.653 ---------------------------------------------------------------------------------------------------------------- *Significant at.05 level; **Significant at.01 level; ***Significant at.001 level. ns =non-significant when entered into model 19

20. 20 Summary of Results Variables with the largest beta coefficient: BusinessStarting salary of graduates EducationEnrollment size EngineeringEnrollment size LawAdmissions selectivity (LSAT) MedicineAdmissions selectivity (MCAT)

21. 21 Summary of Results The SIZE variable (full-time enrollment) is the only variable that remained significant in the final model for all 5 disciplines. However, size has the greatest beta coefficient in only 2 of the 5 disciplines (education and engineering). So for schools of education and engineering, enrollment size is the strongest predictor of reputation!

22. Summary of Results ADMISSIONS SELECTIVITY (average entering test score) remains significant in the final model for 4 of the 5 disciplines, and has the greatest beta coefficient for 2 of those 4 – Law schools and Med schools. So for Law schools and Med schools, the tested “quality” of the admitted students is the strongest predictor of reputation! Education is the one discipline for which the Admissions Test score is not signif. related to reputation. 22

23. Summary of Results FACULTY PRODUCTIVITY (pubs per faculty) also remained significant in 4 of the 5 disciplines, and had the 2 nd greatest beta coefficient in all 4. The 4 disciplines were: engineering, education, law, and medicine. Not surprising that faculty productivity is significant in explaining graduate reputation. 23

24. Summary of Results The business schools may be the most surprising all around --- not only is it the one discipline in which faculty productivity does not influence reputation, but the factor with the greatest influence on reputation is the starting salary of the graduates …..a factor determined by external (market) forces!! 24

25. Summary of Results TUITION (our only measure of wealth) did not remain significant in the final model for any of the 5 disciplines. Student-faculty ratio only remained significant in 2 of the 5 disciplines (Law and Med), and was one of the weaker predictors even for them. 25

26. 26 Implications These results confirm prior studies on graduate reputation that analyzed the 1995 NRC ratings, as well as findings that analyze the correlates of institutional reputation at the UG level. The question remains as to how well the U.S. News U.S. News ratings measure the concept of quality. Is the magazine really determining Best “America’s Best Graduate Schools?”