Cohort Analysis of Retention, Promotion and Tenure for Tenured/Tenure track Faculty, Texas A&M University: Executive Summary

Cohort Analysis of Retention, Promotion and Tenure for Tenured/Tenure track Faculty, Texas A&M University: Executive Summary This report summarizes findings from the fall 2015 analysis of retention and promotion among tenured and tenure-track faculty members. Because those constructs are tightly linked at Texas A&M University, any analysis of retention among assistant professors also incorporates an analysis of tenure. This analysis covers the 10 most recent faculty cohorts (Fiscal Year 2006 through Fiscal Year 2015) at each rank: assistant professor, associate professor, and professor. Thus, analysis of tenure-track faculty examines all faculty members hired at or promoted to the rank of assistant professor since 2006; analysis of associate professors examines all faculty members hired at or promoted to the rank of associate professor since 2006; and analysis of professors examines all faculty members hired at or promoted to the rank of professor since FY 2006. Faculty members who were hired before 2006 were only included in the analysis if they were promoted into one of the cohorts during the analysis period (2006-2015). Because promotion decisions occur infrequently in the life of a faculty member and many departments have years where no faculty members are promoted or separated, we could not analyze promotion and retention at the department level, but we were able to examine differences across colleges and between STEM and non-stem departments within colleges. The analysis examines three key indicators for systematic differences by sex and STEM status: (i) Cohort promotion rates (for assistant and associate professors), which provide information about percentages of faculty from the 10 cohorts who have been promoted at some point during the analysis period. (ii) Cohort separation rates (for associate professors), which mirror the cohort promotion analysis, and are used to understand percentages of faculty from the 10 cohorts who left the University as tenured/tenure-track faculty members. (iii) Eight-year survival rates (for all three ranks), which reflect cumulative effects of annual retention and illustrate demographically-adjusted patterns of persistence among the members of the designated cohorts. Regression analysis has been used to adjust for differences in College, sex, race/ethnicity, national origin, years since the highest degree was granted, time in rank and the annual average rate of growth in faculty salaries at TAMU. Key Findings Regarding Assistant Professors STEM Departments: There were 78 female faculty members and 189 male faculty members hired at or promoted to the rank of assistant professor during the analysis 1

period. Of those 78 female assistant professors, 31% left the university, 27% were promoted to the rank of associate professor and the rest remain at the rank of assistant professor. Similarly, 22% of the 189 male assistant professors left the university while 45% were promoted. Non-STEM Departments: There were 94 female faculty members and 95 male faculty members hired at or promoted to the rank of assistant professor during the analysis period. Of those 94 female assistant professors, 28% left the university, 30% were promoted to the rank of associate professor and the rest remain at the rank of assistant professor. Similarly, 25% of the 95 male assistant professors left the university while 35% were promoted. In prior reports, tenure-track retention (which implies promotion to the rank of associate professor) was significantly lower for female faculty than for male faculty in the Dwight Look College of Engineering, and significantly higher for female faculty than for male faculty in the STEM departments of the College of Agriculture and Life Sciences (COALS), all other things being equal. Within these 10 cohorts, there are no differences in male and female retention rates that are statistically significant at the 5% level. The gap in retention rates has narrowed somewhat, but importantly, the size of the cohort has shrunk, leading to larger standard errors and less precise estimates. o Although the differences are only statistically significant at the 10% level, the predicted eight-year survival rates are more than 19 percentage points lower for females than for males in the Dwight Look College of Engineering, the non- STEM departments of the College of Liberal Arts and the Mays School of Business. o There is also a 19 percentage point difference between male and female eight-year survival rates in the College of Architecture, but that difference is not statistically significant at even the 10% level. Predicted eight-year survival rates for STEM assistant professors are lowest in the College of Geosciences for both male and female faculty; eight-year survival rates for non-stem assistant professors are lowest in the Mays School of Business for both male and female faculty. o The eight-year survival rate for female assistant professors in the Mays School of Business is less than 30%. Key Findings Regarding Associate Professors STEM Departments: There were 95 female faculty members and 313 male STEM faculty members hired at or promoted to the rank of associate professor during the analysis period. Of those 95 female associate professors, 18% left the university and 21% were promoted to the rank of full professor. Male STEM associate professors were more than twice as likely to be promoted (32%) as to leave the university (12%). 2

Non-STEM Departments: There were 114 female faculty members and 149 male STEM faculty members hired at or promoted to the rank of associate professor during the analysis period. Of those 114 female associate professors, 18% left the university and 19% were promoted to the rank of full professor. Similarly, 20% of the 149 male associate professors left the university while 23% were promoted. There is no evidence that female faculty members are more likely than male faculty to leave shortly after earning tenure. For both sexes, the probability of turnover is highest in the first few years after someone earns tenure and promotion. The probability of turnover becomes very small after approximately five years. Male associate professors in the College of Education were significantly more likely to be promoted to the rank of full professor than were female associate professors. No other College had a significant difference in promotion rates between male and female faculty, once other demographic characteristics (such as time-in-rank or years since graduation) are taken into account. Despite apparent differences, there were no Colleges where eight-year survival rates were systematically different for male and female associate professors at even the 10% level. Key Findings Regarding Professors Female faculty comprise less than one sixth of the 293 STEM faculty members (46 females and 247 males) hired at or promoted to the rank of full professor between 2006 and 2015. In contrast, females comprise one third of the 184 non-stem faculty members (64 females and 120 males) in the full professor cohorts. Three Colleges (Geosciences, Architecture and Mays) continue to retain all of the female members from these 10 cohorts. o All of the female members from the three Colleges with 0% separation rates tallies only 13 individuals. The non-stem COALS departments continue to retain all 20 male professors from these 10 cohorts. Across the university, predicted eight-year survival rates are lowest for newly-minted male full professors in the Mays School of Business and for newly-minted female full professors in the Dwight Look College of Engineering, all other things being equal. 3

Cohort Analysis of Retention, Promotion and Tenure for Tenured/Tenure track Faculty, Texas A&M University By Lori L. Taylor and Jeff Froyd November 2015 Promotion and retention are key concerns for any university. Furthermore, at Texas A&M University, promotion and tenure are closely linked. With very few exceptions, faculty members either arrive with tenure, or are granted tenure when they are promoted to the rank of associate professor. Therefore, by examining the extent of differences between the sexes in the promotion and retention of tenure-track faculty, we are simultaneously examining differences with respect to the granting of tenure. Using regression analysis and the most recent 10 cohorts of faculty (Fiscal Year 2006 through Fiscal Year 2015), we examined promotion and retention of faculty members at the assistant and associate professor ranks, and the retention of faculty at the full professor rank. The analyses for the assistant professor cohorts (i.e., those faculty hired at the rank of assistant professor at any time since 2006) and the full professor cohorts (i.e. those faculty hired at or promoted to the rank of full professor at any time since 2006) rely on a Cox proportional hazards model with two possible outcomes retention or separation. The analysis for the associate professor cohorts (i.e., all faculty hired at or promoted to the rank of associate professor at any time since 2006) includes not only the Cox proportional hazards model (to evaluate turnover) but also a multinomial logit model with three outcomes retention at rank, promotion or separation. In all three cases, we pooled multiple cohorts of faculty to enhance the size of the analysis sample. All analyses also controlled for the following factors: college, sex, race/ethnicity, national origin, years since the highest degree was granted, time in rank and the annual average rate of growth in faculty salaries at TAMU. 1 Because promotion decisions occur infrequently in the life of a faculty member and many departments have years where no faculty members are promoted or separated, we could not analyze promotion and retention at the department level, but we were able to examine differences across colleges and between STEM and non-stem departments within colleges. Analyzing the Assistant Professor Cohorts Table 1 reports promotion and separation rates for the assistant professor cohorts. As the table illustrates there were 78 female STEM faculty members and 189 male STEM faculty members hired at or promoted to the rank of assistant professor during the analysis period (2006 through 2015). Of those 78 female assistant professors, 31% left the university, 27% were promoted to 1 Model coefficients and standard errors are available upon request. The Bush School of Government is not included in this analysis because during the early years of this analysis, there were too few faculty members for reliable inference. In all cases, statistical significance is determined at the 5% level. 4

the rank of associate professor and the rest remain at the rank of assistant professor. Similarly, 22% of the 189 male assistant professors left the university while 45% were promoted. Among non-stem colleges and departments, 28% of the female assistant professors and 25% of the male assistant professors left the university and somewhat higher percentages (30% for females and 35% for males) were promoted. Table 1 Career Promotion and Separation Rates for Assistant Professors Hired Since 2006, by College, STEM Status and Sex Number of Individuals Separation Rate Promotion Rate Female Male Female Male Female Male STEM Departments Agriculture and Life Sciences 13 29 7.7% 13.8% 53.9% 58.6% Engineering 24 77 45.8% 19.5% 25.0% 44.2% Geosciences 6 19 33.3% 36.8% 16.7% 31.6% Liberal Arts 29 30 34.5% 36.7% 10.3% 40.0% Science 6 34 0.0% 11.8% 66.7% 47.1% Total STEM 78 189 30.8% 21.7% 26.9% 45.0% Agriculture and Life Sciences 6 18 16.7% 33.3% 0.0% 38.9% Architecture 12 19 25.0% 31.6% 33.3% 26.3% Education 14 7 21.4% 28.6% 64.3% 14.3% Liberal Arts 32 19 31.3% 21.1% 31.3% 47.4% Mays School of Business 12 23 58.3% 17.4% 0.0% 39.1% Veterinary Medicine 18 9 11.1% 22.2% 27.8% 22.2% Total Non-STEM 94 95 27.7% 25.3% 29.8% 34.7% Total STEM and Non-STEM 172 284 29.1% 22.9% 28.5% 41.6% Note: The table includes individuals who were hired between 2006 and 2015. Most who were neither promoted nor separated continue to make satisfactory progress toward tenure. Of course, there are differences across the Colleges and between individual faculty members with respect to time in rank, years since degree, etcetera. The Cox proportional hazards model allows us to adjust for demographic differences in the data. Previous analyses (such as the analysis of the 2004-2013 period) found that tenure-track retention (which implies promotion to the rank of associate professor) was significantly lower for female faculty than for male faculty in the Dwight Look College of Engineering, and significantly higher for female faculty than for male faculty in the STEM departments of the College of Agriculture and Life Sciences (COALS), all other things being equal. Across these cohorts, the gaps in retention rates have 5

narrowed somewhat and the cohorts are much smaller. 2 As a result, there are no differences in male and female retention rates that are statistically significant at the 5% level (the significance standard used in prior reports). Table 2 compares the estimated eight-year survival rates for assistant professors across the various divisions at TAMU. As the table illustrates, predicted survival rates for STEM assistant professors are highest in the College of Agriculture and Life Sciences for male faculty, and in the College of Science for female faculty. Survival rates for STEM assistant professors are lowest in the College of Geosciences for both male and female faculty. Survival rates for non-stem assistant professors are lowest in the Mays School of Business for both male and female faculty. Survival rates are more than 19 percentage points higher for males than for females in the Colleges of Engineering and Architecture, the non-stem departments of the College of Liberal Arts and the Mays School of Business. With the exception of the College of Architecture, all of these differences are statistically significant at the 10% level. Table 2 Eight-Year Survival Rates for Tenure Track Faculty by College, STEM Status and Sex Female Male STEM Departments Agriculture and Life Sciences 73.1% 78.3% Engineering 49.0%* 68.7% Geosciences 46.9% 41.1% Liberal Arts 55.9% 49.2% Science 76.1% 69.2% Agriculture and Life Sciences 69.1% 52.4% Architecture 54.5% 74.4% Education 58.1% 52.2% Liberal Arts 56.8%* 76.4% Mays School of Business 29.7%* 50.0% Veterinary Medicine 78.0% 74.1% Note: The analysis includes individuals who were hired at or promoted to the rank of Assistant Professor any year during the analysis period (2006-2015). Survival rates calculated holding constant race, sex, national origin, years since degree, STEM status and College/Division. The asterisks indicate that the difference between males and females is statistically significant at the *** 1% level, **5% level or *10%level. 2 The 2004 2013 assistant professor cohorts included many faculty reinvestment hires. These 10 cohorts had a combined 100 STEM female and 228 STEM male assistant professors, or 23% more faculty members than the 2006 2015 cohorts. Similarly, the 2004 2013 non STEM cohorts had 106 female and 122 male assistant professors, or 21% more faculty members than the 2006 2015 cohorts. 6

Analyzing the Associate Professor Cohorts Table 3 reports promotion and separation rates for the associate professor cohorts. As the table illustrates there were 95 female STEM faculty members and 313 male STEM faculty members hired at or promoted to the rank of associate professor during the decade from 2006-2015. Of those 95 female associate professors, 18% left the university and 21% were promoted to the rank of full professor; the rest remain at the rank of associate professor. Similarly, 12% of the 313 male associate professors left the university while 32% were promoted. Among the non-stem departments, female associate professors were equally likely to leave the university (18%) as to be promoted to full professor (19%), while male associate professors in non-stem departments were slightly more likely to be promoted (23%) than to leave (20%). Female associate professors were most likely to be promoted to the rank of full professor in the College of Geosciences, and least likely to be promoted in the non-stem departments of the College of Agriculture and Life Sciences, but neither College had a large number of female associate professors, so the apparent difference may not be statistically meaningful. Table 3 Cohort Promotion and Separation Rates for Associate Professors Hired or Promoted Since 2006, by College, STEM Status and Sex Number of Individuals Separation Rate Promotion Rate Female Male Female Male Female Male STEM departments Agriculture and Life Sciences 21 56 14.3% 12.5% 19.1% 25.0% Engineering 24 134 20.8% 11.9% 16.7% 24.6% Geosciences 8 27 12.5% 14.8% 37.5% 44.4% Liberal Arts 25 30 24.0% 10.0% 16.0% 30.0% Science 17 66 11.8% 12.1% 29.4% 47.0% Total STEM 95 313 18% 12% 21% 32% Agriculture and Life Sciences 10 18 10.0% 16.7% 10.0% 27.8% Architecture 12 20 8.3% 25.0% 16.7% 10.0% Education 33 33 27.3% 21.2% 18.2% 33.3% Liberal Arts 33 32 12.1% 15.6% 18.2% 15.6% Mays School of Business 9 23 33.3% 26.1% 22.2% 17.4% Veterinary Medicine 17 23 17.7$ 17.4% 29.4% 34.8% Total Non-STEM 114 149 18% 20% 19% 23% Total STEM and Non-STEM 209 462 18% 15% 20% 29% Note: The table includes individuals who were hired at or promoted to the rank of associate professor at any time during the analysis period (2006-2015). 7

Regression analysis allows us to control for demographic differences among faculty members and test for statistical significance. It indicates that with the exception of the College of Education (where male promotion rates were significantly higher than female promotion rates) none of the apparent differences in promotion rates between the male and female faculty within a college are statistically reliable, once other demographic characteristics (such as time-in-rank or years since graduation) are taken into account. There is no evidence that differences in separation rates between male and female faculty within any college are statistically significant. The small number of female faculty in some colleges may be skewing the within-college results, however. Therefore, we pooled the data across colleges and examined the possibility that there could be systematic differences between male and female faculty that were common across the university. Figure 1: Demographically Adjusted Probabilities of Promotion and Separation for Associate Professors, 2006-2015 0.05.1.15.2.25 Probabilities of Promotion and Separation All Divisions 0 2 4 6 8 10 Years as TAMU Associate Professor Female Probability of Promotion Female Probability of Separation Male Probability of Promotion Probability of Separation Figure 1 illustrates the probability of separation and promotion for male and female faculty, holding all other demographic characteristics constant at the mean and assuming that the faculty member was not hired in at the rank of associate professor (and indeed, very few women were). As the figure indicates, the predicted probability of promotion was slightly higher for a male 8

associate professor with fewer than seven years of time in rank than it was for a female associate professor with the same demographic characteristics. However, the difference is not statistically significant Figure 1 also provides insight into the pattern of retention among associate professors. As the figure illustrates, turnover among associate professors is highest in the first few years after they earn tenure and promotion. The probability of turnover becomes very small after approximately five years. There is no evidence that female faculty are more likely than male faculty to leave shortly after earning tenure. Rather, the pattern of recent history suggests that associate professors of both sexes are least likely to stay when they are in the first few years post tenure. Table 4 presents results from an analysis of turnover among associate professors. This analysis focuses exclusively on faculty retention and therefore does not distinguish between retention in rank and retention after promotion. As the table illustrates, there are few systematic differences in eight-year survival rates between male and female associate professors. None of the apparent differences in survival rates were statistically significant. Table 4 Eight-Year Survival Rates for Associate Professors Hired or Promoted Since 2004 by College, STEM Status and Sex Female Male STEM Departments Agriculture and Life Sciences 77.3% 75.1% Engineering 66.3% 79.4% Geosciences 81.3% 81.0% Liberal Arts 66.1% 80.4% Science 71.4% 72.0% Agriculture and Life Sciences 85.0% 73.3% Architecture 80.2% 56.4% Education 62.0% 69.3% Liberal Arts 74.8% 77.8% Mays School of Business 63.3% 44.7% Veterinary Medicine 69.7% 69.1% Note: The analysis includes individuals who were hired at or promoted to the rank of Associate Professor any year during the analysis period (2006-2015). Survival rates calculated holding constant race, sex, national origin, age group, rank at hire, years since degree, educational attainment, administrator status, STEM status and College/Division. The asterisks indicate that the difference between males and females is statistically significant at the *** 1% level, **5% level or *10%level. Analyzing the Full Professor Cohorts Table 5 presents separation rates for the last 10 full professor cohorts. As the table illustrates, female faculty comprise slightly more than one third of the non-stem but less than less than one sixth of the STEM faculty members hired at or promoted to the rank of full professor between 9

2006 and 2015. Separation rates are slightly lower for female full professors than for male full professors, on average. Three Colleges (Geosciences, Architecture and Mays) continue to retain all of the female members from these 10 cohorts. However, all of the female members from these 10 cohorts tallies only 13 individuals. Table 5 Cohort Separation Rates for Full Professors Hired or Promoted Since 2006 by College, STEM Status and Sex Number of Individuals Separation Rate Female Male Female Male STEM departments Agriculture and Life Sciences 8 39 12.5% 18.0% Engineering 11 92 18.2% 13.0% Geosciences 4 30 0% 16.7% Liberal Arts 10 21 10.0% 28.6% Science 13 65 15.4% 20.0% Total STEM 46 247 13% 17% Agriculture and Life Sciences 6 20 16.7% 0% Architecture 5 14 0% 28.6% Education 16 20 18.8% 10.0% Liberal Arts 21 25 19.1% 8.0% Mays School of Business 4 21 0% 42.9% Veterinary Medicine 12 20 16.7% 15.0% Total Non-STEM 64 120 16% 17% Total STEM and Non-STEM 110 367 15% 17% Note: The table includes individuals who were hired at or promoted to the rank of Professor any year during the analysis period (2006-2015). Again, there are differences across the Colleges and among individual faculty members with respect to time in rank, years since degree, etc. As with the analyses of the assistant professor cohorts, the Cox proportional hazards model allows us to adjust for demographic differences in the data. Not surprisingly, it indicates that retention has been significantly higher for female faculty than for male faculty in the three Colleges where no female full professors from these 10 cohorts have turned over, and significantly higher for male faculty than for female faculty in the one College where no male full professors from these 10 cohorts have turned over (COALS non- STEM). 10

Table 6 compares the eight-year survival rates for professors across the various divisions at TAMU. As the table illustrates, across the university, predicted eight-year survival rates are lowest for newly-minted male full professors in the Mays School of Business and for newlyminted female full professors in the Dwight Look College of Engineering, all other things being equal. Table 6 Eight-Year Survival Rates for Full Professors Hired or Promoted Since 2004 by College, STEM Status and Sex Female Male STEM Departments Agriculture and Life Sciences 81.2% 71.0% Engineering 70.8% 80.5% Geosciences 100.0%*** 78.3% Liberal Arts 76.2% 52.3% Science 78.4% 79.1% Agriculture and Life Sciences 80.2%*** 100.0% Architecture 100.0%*** 74.8% Education 83.0% 85.8% Liberal Arts 76.7% 89.0% Mays School of Business 100.0%*** 42.0% Veterinary Medicine 86.7% 83.0% Note: The analysis includes individuals who were hired at or promoted to the rank of Professor any year during the analysis period (2006-2015). Survival rates calculated holding constant race, sex, national origin, age group, years since degree, rank at hire, educational attainment, administrator status, STEM status, and College/Division. The asterisks indicate that the difference between males and females is statistically significant at the *** 1% level, **5% level or *10% level. 11