An Analysis of Faculty Turnover at the University of Wisconsin -Madison
Margaret N. HarriganUniversity of Wisconsin – Madison
Office of Budget, Planning and Analysis Presented at the 39th Annual AIR Forum, May 30 – June 2, 1999, Seattle, Washington
I. Introduction
This paper examines the rate of faculty turnover at the University of Wisconsin – Madison. Overall faculty retention rates and the rate of turnover for tenured and probationary faculty are described. An earlier study at UW-Madison found significant differences in the retention rates of women and men hired with probationary status. Therefore, the paper also focuses on the differences in turnover rates for men and women faculty and describes some of the policies adopted to improve faculty retention.
A certain amount of turnover is necessary and healthy for an institution. Retirements and other terminations can create opportunities for a change in focus of academic programs or development or expansion of new curricular areas. However, too much staff turnover can lead to excessive costs. In addition to the costs of the actual search, there are often costs associated with hiring temporary replacements until a faculty position can be filled. Particularly at a research institution, the costs of hiring a new faculty member can exceed a half million dollars for laboratory equipment, space, and funding for graduate assistants as part of a start-up package. Higher turnover rates for women or minority faculty may signal problems within the university, including issues of overall climate, equity in salaries, or problems within the tenure process.
II. How much is "normal" turnover?
If all faculty were hired and retained until retirement after 30 years of service, we would expect an equilibrium turnover rate of about one-third of the faculty every ten years or 3.3 percent per year. An alternative hypothetical university, which hired all its faculty on probation and which denied tenure to all of them in their seventh year, would have an equilibrium turnover rate of 1/7 or 14.3 percent per year. Thus, we would expect the "normal" turnover rate to fall somewhere between these two extremes.
The University of Wisconsin – Madison currently has 2135 faculty members, down from a high of over 2400 from 1985 to 1990. In the past few years, the number of retirements and deaths has increased to about 80 per year. In addition, about 50 faculty per year leave for reasons other than retirement or death. About 60-90 new faculty have been hired in each of the past five years. Based on a simple year-over-year comparison for recent years, it appears that UW-Madison’s turnover rate is about 5-6 percent per year.
A brief literature search and an e-mail survey of other research universities yielded some information on faculty attrition at other institutions. Most analyses reported only the turnover rate after one year. These ranged from about two percent to ten percent and are reported in Table 1 (University of Alaska 1997, Cornell University 1998, South Dakota Board of Regents 1997 and 1998, Florida 1994, Brown and Woodbury 1995, Harrigan 1998). In addition, the University of Alaska also reported five-year retention rates – about 75 percent of the faculty were still employed after five years. An analysis of AAUP data computed retention rates separately by rank, and reported retention rates of 90-92 percent for full and associate professors and 84-86 percent for assistant professors (Ehrenberg et al. 1991).
Faculty Turnover After One Year
| College or University |
Percent of Faculty Who Left in One Year | Year Analyzed |
| University of Alaska | 1.7% 6.1 6.9 |
1996 1995 1994 |
| Cornell University | 4.9 6.1 4.0 |
1996-97 1995-96 1994-95 |
| University of Florida | 7 | 1989-90 |
| University of Iowa | 4 4 3 |
1996-97 1995-96 1994-95 |
| University of Kansas | 4.3 3.9 4.5 |
1996-97 1995-96 1994-95 |
| University of Michigan | 3.6 | 1991-97 |
| Michigan State University | 4.5 3.5 3.3 3.7 4.8 |
1996-97 1995-96 1994-95 1993-94 1989-90 |
| University of Minnesota | 5.4 3.8 6.7 4.4 3.9 |
1996-97 1995-96 1994-95 1993-94 1992-93 |
| Occidental College | 1.7 | 1980-94 |
| Ohio State University | 5.5 3.2 |
1996-97 1995-96 |
| University of South Dakota | 7.9 10.1 |
1996-97 1995-96 |
| South Dakota State University | 5.9 6.2 |
1996-97 1995-96 |
| University of Wisconsin-Madison | 5.9 5.1 6.4 5.8 |
1997-98 1996-97 1995-96 1994-95 |
SOURCES: University of Alaska 1997, Cornell University 1998, Honeyman and Summers 1994, Brown and Woodbury 1995, Tamada and Inman 1997, South Dakota Board of Regents 1998 and 1997, Harrigan 1998.
[TOP/Table of Contents] [I. Introduction] [II. How much is "normal" turnover?] [III. Faculty Retention After Ten Years][ IV. Cohort Analysis] [V. Policies to Reduce Turnover] [VI. Conclusions] [ VII. References] [VII. Footnotes]
[List of Tables and Charts]
III. Faculty Retention After Ten Years
If the size of the faculty at an institution is significantly expanding or contracting over time, a simple year-over-year comparison is misleading. For example, an early retirement incentive program effective in one year will likely lead to artificially high turnover in that year followed by a reduced rate in the following year. A more informative approach to analyzing faculty turnover takes a single group of faculty at one point in time and follows them for a specified period. Therefore, I grouped UW-Madison faculty into several cohorts to analyze turnover trends. For each group, a fixed number of faculty are analyzed for a ten-year period. For example, I examined all faculty who were employed in 1978-79 to determine whether they were still employed ten years later. I repeated this process for all faculty who were employed in 1979-80 and each year for which I had ten years of data (through 1987-88). (Note that an individual who was hired prior to 1978 and retained for more than ten years will be included in each of the cohorts shown.) Table 2 shows the results of this analysis. For faculty of all ranks, the average retention rate after ten years for the period analyzed is about 55 percent.
Faculty Retention Rates at UW-Madison after Ten Years
| 78-79 | 79-80 | 80-81 | 81-82 | 82-83 | 83-84 | 84-85 | 85-86 | 86-87 | 87-88 | |
| Total | 2384 | 2349 | 2337 | 2323 | 2330 | 2366 | 2388 | 2466 | 2437 | 2415 |
| Women | 323 | 324 | 327 | 329 | 331 | 342 | 358 | 384 | 376 | 367 |
| Men | 2061 | 2025 | 2010 | 1994 | 1999 | 2024 | 2030 | 2082 | 2061 | 2048 |
| Percent Still Here | ||||||||||
| Women | 42.7% | 44.4% | 43.1% | 45.0% | 45.0% | 46.5% | 46.6% | 49.0% | 51.9% | 56.7% |
| Men | 59.2% | 58.4% | 57.4% | 57.0% | 57.2% | 56.5% | 55.6% | 52.4% | 51.6% | 54.0% |
| Percent who retired or died | ||||||||||
| Women | 20.1% | 19.8% | 22.0% | 22.2% | 22.4% | 20.5% | 21.2% | 21.6% | 21.5% | 20.4% |
| Men | 19.9% | 21.8% | 23.3% | 25.0% | 25.2% | 26.0% | 28.0% | 31.7% | 33.1% | 31.8% |
| Percent who left for other reasons | ||||||||||
| Women | 37.2% | 35.8% | 34.9% | 32.8% | 32.6% | 33.0% | 32.1% | 29.4% | 26.6% | 22.9% |
| Men | 20.9% | 19.8% | 19.3% | 18.0% | 17.6% | 17.4% | 16.5% | 15.8% | 15.3% | 14.2% |
SOURCE:UW-Madison tenure file and IADS personnel file (September 1997), Prepared by: Office of Budget Planning and Analysis (MH)
A comparison of the percent of men and women faculty who were retained for ten or more years is also shown in Chart 1. The retention rate of male faculty has declined slightly in the period studied, from 59 percent in the 1978-79 cohort to 54 percent in the 1987-88 cohort. The proportion of women faculty who were still here after ten years improved over the analysis period, from about 43 percent in the earliest cohort studied to 57 percent in the latest. For faculty who were here in the 1985-86 and later cohorts, there is essentially no difference in retention rates for men and women.
However, this analysis masks several differences between men and women faculty. Chart 2 compares the proportion of faculty leaving due to death or retirement. In the early cohorts, men and women death/retirement rates were very similar. Male faculty here in 1985-86 and later cohorts report retirement rates about ten percent higher than women. This is likely due to in part to changing age distributions of the faculty. Over the time period shown in Table 2, the number of women faculty increased by about 13 percent, to 367 in 1987, while the number of male faculty fell by one percent. Women faculty, a higher percentage of which are new hires, are more likely to be younger and less likely to have tenure than men faculty. For example, currently about seven percent of women faculty are at least 60 years of age; 19 percent of male faculty are 60 or older.
The proportion of faculty who resigned for reasons other than death or retirement is graphed in Chart 3. For both men and women, the percent who left for other reasons declined over the analysis period. The decrease is most dramatic for women faculty - from 37.2 percent in 1978-79 to 22.9 percent in 1987-88.
Chart 1, Chart 2 and Chart 3 follow:
NOTES: See TABLE 2.
SOURCE: UW-Madison tenure file and IADS personnel file (September 1997)
NOTES: See TABLE 2.
SOURCE: UW-Madison tenure file and IADS personnel file (September 1997), NOTES: See TABLE 2.
[TOP/Table of Contents] [I. Introduction] [II. How much is "normal" turnover?] [III. Faculty Retention After Ten Years][ IV. Cohort Analysis] [V. Policies to Reduce Turnover] [VI. Conclusions] [ VII. References] [VII. Footnotes]
[List of Tables and Charts]
IV. Cohort Analysis
The reasons faculty members leave university employment can be grouped into three general categories: involuntary (fiscal emergency, did not earn tenure, dismissed for cause); voluntary (dissatisfied with position, found better career opportunity, higher salary elsewhere); or end of career (retirement or death). The likelihood of each reason varies depending on the stage of one's career. Once a faculty member has achieved tenure, involuntary dismissal is quite rare. However, a significant portion of probationary faculty leaves because they have been denied tenure or believe they will be denied tenure. A survey of UW Madison probationary faculty hired between 1977-80 indicated that about 60 percent of the men and 50 percent of the women who left believed that they would be denied tenure (Reed et al. 1988). Thus, one would expect the turnover rates for tenured and untenured faculty to differ. The age distribution of faculty will also influence turnover. Probationary faculty are typically younger and much less likely to retire than tenured professors. In 1998, of 2135 faculty, 28 percent were over 55 years of age (two of these were untenured assistant professors). Therefore, I have divided the faculty into two analysis groups: probationary and tenured.
The University of Wisconsin – Madison established a database in 1978 to track the appointments of tenured and probationary faculty. It includes the date of hire, status (tenured or probationary), department(s) where the faculty member holds tenure or tenure-track status, date probation began, date tenure granted, date of promotion to full professor, date appointment ended for those who left, and other faculty characteristics (1). This database allowed me to follow the employment status of 2191 faculty hired between June 1978 and May 1998 (2). Of these, 80 percent were hired as probationary appointments.
- Other Universities
- UW-Madison
A few other universities have compiled data on the turnover rate of probationary faculty. Miami University of Ohio (Coalition on the Status of Women Faculty 1997) analyzed faculty hired from 1982 to 1990 and reported an overall successful tenure rate of 56 percent for women and 70 percent for men. At the University of Missouri, tenure-track faculty hired between 1982 and 1986 were analyzed (Eimers 1995). About 45 percent earned tenure by the end of the seventh year (three percent were still in their probationary appointment). No significant difference in tenure rates was found between men and women. Ohio State University examined the attrition rate of untenured faculty hired between 1987-88 and 1992-93 (Harrigan 1998). By June 1997, 32 percent of Arts and Sciences faculty, 41 percent of those in the professional colleges, and 36 percent of health sciences faculty had left.
Two published studies analyzed a subset of probationary faculty at their institutions. At one university, retention of tenured and tenure-track faculty in large social science departments was analyzed (Schwab 1991). Of 58 probationary faculty on the payroll in 1980-81, 56 percent had left the university by 1986-87. A study at the University of Minnesota examined the promotion rate of probationary faculty hired between 1975 and 1985 in 13 hard science departments (Kingsbury Jones et al. 1994). Of the 104 faculty (of whom 20 were women), 72 percent earned tenure. Women were slightly more likely to earn tenure than men (85 percent and 69 percent, respectively), but the difference was not significant.
[TOP/Table of Contents] [I. Introduction] [II. How much is "normal" turnover?] [III. Faculty Retention After Ten Years][ IV. Cohort Analysis] [V. Policies to Reduce Turnover] [VI. Conclusions] [ VII. References] [VII. Footnotes]
[List of Tables and Charts]
UW-Madison is particularly interested in the retention rate of its probationary faculty since 80 percent of the faculty are hired as instructors or assistant professors. In addition, an earlier UW study found a large gap in the retention rates of men and women assistant professors (Reed et al. 1988). The current study was undertaken, in part, to determine whether retention rates for men and women had improved since then.
At UW-Madison, the typical faculty member is promoted to tenure during the fifth or sixth year of his or her probationary period. Nine percent are promoted in less than five years (3). About six percent of those promoted in the past decade had a probationary period of more than seven years. In Table 3, the proportion of probationary faculty who were promoted within seven years and nine years is presented (including time for prior service at other universities).
At UW-Madison, 1467 probationary faculty (including 477 women) were hired between July 1978 and June 1993. Of these, about 58 percent received tenure within seven years and about 2 percent were still employed as probationary faculty. Women were less likely to receive tenure than men – 47.6 percent of women faculty and 62.4 percent of men faculty were promoted within seven years. Seven percent of women and two percent of men faculty were still in probationary appointments at the end of seven years. The gap in promotion rates narrows slightly when a nine-year probationary period is studied: 51.4 percent of women and 63.8 percent of men earned tenure in that time period.
The proportion of men who earned tenure varied somewhat across the period studied, ranging from 57 to 70 percent. Considerably more variation is observed in the proportion of women who earned tenure. Less than 30 percent of women hired in 1978 were promoted; 87 percent of those who started in 1986 earned tenure. In five of the cohorts studied (all before 1985), less than half of the women were promoted.
Notes: Includes 1467 faculty hired with probationary status from June 1978 through May 1993. No adjustment made for extensions to the tenure clock; includes credit for prior service at other universities. Three women faculty who were promoted after nine years are shown here as not promoted.
Source: UW-Madison Tenure Tracking data base and Integrated Appointment Data System data base (5/99)
- Analysis by Discipline Area
- Time to Promotion
(Chart 4)
NOTES: See Table 3.
At UW-Madison, faculty are divided into four broad discipline areas for the purposes of tenure review: humanities, social studies, physical sciences, and biological sciences. In addition to departmental executive committees, divisional committees made up of faculty across several colleges must vote on each tenure case. The following section examines the tenure rate of faculty by discipline area.
Women faculty at UW are predominantly in the humanities and social studies fields. In contrast, over 60 percent of the men are in the physical and biological science areas. If the tenure rate is higher for faculty in the natural sciences than in social studies and humanities, the difference in promotion rates could be due to discipline field rather than gender.
Table 4 presents data on the tenure rate of men and women by the four disciplinary areas. Men and women faculty in the social studies area are equally likely to receive tenure. Similarly, there is little difference between the promotion rates of men and women in the physical sciences since 1982. However, an individual hired in a physical sciences department is much more likely to earn tenure than one in social studies -- about 70 percent of the former and half of the latter were promoted. In the biological sciences, women are less likely than men to be promoted. About 44 percent of women and 68 percent of men in biological sciences were granted tenure in the period analyzed. If women were in the same discipline areas as men, the overall tenure rate for women would increase by about two percent. Thus, most of the gap in tenure rates between men and women is not explained by differences in discipline. (Table 4.)
In three of the discipline areas (humanities, physical sciences, and biological sciences), the tenure rate of women increased over time. The most dramatic improvement was in the physical sciences -- 17 percent of women hired between 1978 and 1982 earned tenure, compared to 88 percent of women hired between 1987 and 1991. However, the number of women hired in the physical sciences is very small. Women in the humanities increased their promotion rate from less than half to nearly two-thirds over the analysis period. The promotion rate for women in the biological sciences improved from 35 percent for those hired in 1978-82 to 55 percent for those hired in the latest period studied.
[TOP/Table of Contents] [I. Introduction] [II. How much is "normal" turnover?] [III. Faculty Retention After Ten Years][ IV. Cohort Analysis] [V. Policies to Reduce Turnover] [VI. Conclusions] [ VII. References] [VII. Footnotes]
[List of Tables and Charts]
The simple cohort analysis used above has limits. Anyone hired since 1990-91 is excluded from much of the analysis, since a nine-year period is not available to measure promotions. UW-Madison has made significant efforts in the past ten years to improve the retention of untenured faculty, particularly women. Restricting the analysis to those hired before 1991 limits our ability to measure the effectiveness of these programs. However, a statistical technique called survival or event history analysis allows one to make use of "censored" observations – that is, individuals for whom the event (e.g., tenure) has not yet occurred. This analytical approach was first applied in biostatistical research to study response to medical treatments, hence the term "survival" analysis.
Survival analysis also allows us to examine the amount of time until the event occurs. Thus we can compare the time to promotion for men and women as well as whether or not tenure is granted. In addition, this approach is used to determine at what point in time an event such as promotion is most at risk of occurring. (See Willett and Singer 1991 and Yaffee and Austin 1994 for discussion and examples of event history analysis in social science research).
For the survival analysis discussed below, 1751 probationary faculty hired between June 1978 and May 1998 are followed for up to 24 semesters (12 years). Table 5 shows the conditional probability of promotion to tenure during each semester for men and women and Chart 5 shows the survival distribution function for men and women. The hazard function, shown in Chart 6, graphically displays the likelihood of tenure occurring for each semester. In this case, we can see that the likelihood of tenure being granted is substantially higher for even-numbered semesters (e.g., at the end of an academic year). This is reasonable, since nearly all promotions at UW take effect following Spring semester. In addition, we find very few tenure decisions being made prior to eight semesters. The time period during which tenure is most likely to be granted for both men and women is after twelve semesters, with slightly lower peaks at 14 and 16 semesters. This reflects the fact that there is a maximum probationary period of seven years, except when an extension to the tenure clock has been granted.
Standard statistical tests can be performed to compare the shapes of the survival curves for men and women. These tests confirm what we see in the charts – the rate of promotion for men and women differ significantly. Virtually no one is promoted with a probationary period of less than six semesters. After that point, the survival functions for men and women diverge. At nearly every point in time, men are at a higher "risk" of tenure. Of those who are promoted, a higher proportion of women than men received tenure after the official seven-year probationary period (4).
(Table 5.)
Chart 5, Chart 6 follow:
NOTES: Includes probationary faculty hired from 1978 through 1998, tracked through May 1999.
In addition to the possibility of being granted tenure, probationary faculty are also at risk of leaving the university without tenure. Table 6 shows the conditional probability of leaving without tenure for men and women; Charts 7 and 8 show the corresponding survival distribution functions and hazard functions. Again, few individuals leave before ten semesters. However, women are more likely to leave in the earlier semesters than are men. The proportion of faculty at risk of leaving increases steadily over time for even-numbered semesters. Few faculty remain after 14 semesters. Because of the maximum probationary period, the likelihood of remaining in probationary status is very low and the probability of tenure or leaving is relatively high after seven years.Again, statistical tests can be used to assess whether the survival functions for staying/ leaving without tenure differ for men and women. The Wilcoxon chi-square test, more sensitive to differences in the earlier time periods, is statistically significant at the 5 percent level, indicating that the rates of leaving without tenure differ for men and women. A test more sensitive at later time periods, the log-rank test shows no statistically significant difference between men and women.
As noted earlier, factors such as year of hire and discipline area are also related to whether or not a person is granted tenure. Many policy changes designed to increase retention were adopted in the late 1980s and early 1990s. Chart 4 above showed very little difference between men and women in the percent tenured within nine years for those hired from 1987-91. Therefore, we might expect the survival function to differ between the earlier and later cohorts. In fact, faculty hired prior to 1987 were significantly more likely to leave without tenure compared to those hired since June 1987. (A survival function shows that after twelve semesters about 31 percent of the earlier cohort and 24 percent of the later cohort have left.)
If only those hired since June 1987 are compared, the survival functions measuring the risk of leaving without tenure are essentially the same for men and women. A survival function examining the difference in rates of promotion over time shows a significant difference for men and women hired since 1987, however. This is likely due to women spending more time in probationary status before being promoted. After ten semesters, the tenure rate is 11 percent for men and 6 percent for women; 19 percent of women and 29 percent of men are promoted within twelve semesters. Women are also more likely to have received an extension to the probationary period. Only a small number of faculty request an extension – about 10-15 per year are granted. A disproportionate number of these are granted to women, however. Of current faculty, about 12 percent of women and 5 percent of men who were hired with probationary status received a tenure clock extension (over half have since received tenure).
Observed differences in the tenure or leaving rates of men and women may be explained by other factors as well, such as discipline area, whether or not the individual had a terminal degree when hired, years of prior experience, or number of publications. A Cox proportional hazard regression model, with explanatory variables for discipline, gender, minority status, years since terminal degree, and hiring cohort was attempted. However, the model failed goodness of fit tests and so is not included here. The Cox proportional hazard model requires that the hazards be proportional. In other words, the graphs of the survival functions should be parallel – the shape of the curves for men and women should be the same, differing only by a constant proportion due to gender. This assumption is violated for the promotion data in this analysis. Furthermore, the model assumes that interaction terms or polynomial forms of the explanatory variables are inappropriate. In fact, such terms appear to improve the fit of a logistic regression model. Additional analyses using discrete-time logistic regression models, with more relaxed assumptions permitted, are in progress.
(Table 6.)
Chart 7, Chart 8 follow:
NOTE: Includes men and women hired in probationary status from 1978 through 1998, tracked through May 1999.
UW-Madison has not previously done a study on the retention rate of tenured
faculty and I found limited comparative studies in my review of the literature. For this
analysis I examined all faculty who were granted tenure at UW from June 1978 through
January 1989. Forty-two women and 255 men were hired with tenure during the period; 497
men and 127 women were promoted to tenure. Table 7 shows the proportion of men and women
who were still on the faculty ten years after tenure. The number of women granted tenure
in any one year is relatively small, averaging about 15 per year over this period.
Therefore, I have collapsed the cohorts into two groups: 1978-83 and 1983-88. The
retention rates for men and women after tenure are essentially equal in this analysis -
73.9 percent of men and 71.6 percent of women were still at UW ten years after tenure was
granted. Of those who left, both men and women were somewhat more likely to resign than to
retire.
(Table 7.)
The percentages of faculty retained are similar in the two time periods. However, faculty are somewhat more likely to retire in the later time period analyzed. This increase in retirements is due in part to an event in 1985. At that time, UW-Madison added about 70 (mostly male) tenured faculty to its roster who transferred from UW Extension. Many of these individuals had been tenured at UW-Extension for a number of years and were more likely to retire than someone who had just earned tenure at UW. If these transferred employes were excluded from the above table, the proportion of faculty in the later cohort who had retired would decrease by about one-third, and the proportion who are retained would increase by about 2-3 percent.
Table 8 divides the tenured faculty into those who were hired with tenure and those who
were promoted to tenure at UW. Men and women had similarly high rates of retention if they
were hired on probation and subsequently granted tenure at UW -- 78 percent of the men and
76 percent of the women were retained after ten years. However, men and women who were
hired with tenure were less likely to stay ten years. Only 57 percent of women and 66
percent of the men faculty who were hired with tenure were retained. Frequently, these
faculty had tenure at another university prior to coming to UW and are likely closer to
retirement age than faculty who were just promoted. As the table shows, 16 percent of the
men and 21 percent of the women retired within ten years. In comparison, retirement rates
of 6 percent and 4 percent, respectively, are observed for men and women who were promoted
to tenure at UW. Faculty hired with tenure may also be more mobile – they have
already left one institution to come to Madison. Because the number of women faculty hired
with tenure is so small, the gap in retention rates between men and women hired with
tenure is not statistically significant.
(Table 8.)
As noted in the section on untenured faculty, the simple cohort analysis described above has limits. Anyone granted tenure since January 1989 is excluded from the analysis, since a ten-year period is not available to measure retention. In the survival analysis discussed below, 1801 faculty newly tenured between June 1978 and May 1998 are included and are tracked for up to 41 semesters (through January 1999). About half of these faculty were hired since January 1989. Chart 9 shows the survival function for tenured men and women faculty over the period analyzed. The hazard function is produced in Chart 10.
Although twenty years of data is available for some individuals, most of the faculty in
this analysis have not yet left the university: 78 percent of the female and 75 percent of
the male observations are censored. Examination of the hazard function indicates that the
risk of leaving is fairly constant through the time period analyzed, with faculty more
likely to leave after even-numbered semesters. (Only about twenty women in this population
were hired before June 1980, so hazard rates calculated at 39 or more semesters are based
on very small numbers.)
(Chart 9, Chart 10 follow:)
NOTE: Includes all UW Madison faculty granted tenure between June 1978 and May 1998, tracked through December 1998.
A comparison of the rates of leaving for tenured faculty shows no statistical
difference between men and women over the period analyzed. However, as expected, the
survival rates for those promoted to tenure differ from those hired with tenure (Chart
11). Those hired with tenure leave have a much higher risk of leaving than do those
promoted to tenure for the time intervals analyzed. This is likely due to the older age of
the faculty hired with tenure. About 40 percent of those promoted to tenure were 40 years
of age or older at the time of promotion. In contrast, over three-fourths of the faculty
hired with tenure were at least fort years of age when hired. Additional analyses, which
control for age at hire, hiring cohort and other variables, may explain some of the
difference in retention between the two groups.
(Chart 11)
NOTE: Includes all UW Madison faculty granted tenure between June 1978 and May 1998,
tracked through December 1998.
[ IV. Cohort Analysis] [V. Policies to Reduce Turnover] [VI. Conclusions] [ VII. References] [VII. Footnotes]
[List of Tables and Charts]
V. Policies to reduce turnover
A number of new policies aimed at improving the retention rate for women faculty have been adopted in recent years. Some are designed to help men faculty as well. Because of the typical length of the tenure process, any changes in hiring policies, mentoring programs, etc. intended to effect the number of promotions will necessarily have a lag before any impacts can be measured. Nonetheless, below are several programs adopted which may influence faculty turnover.
It is often argued that a key to retaining underrepresented employes is to have enough for a "critical mass" to be able to create a sense of community. In 1988, the Madison Plan was established to increase the representation of women in science and minority faculty in all disciplines. Central funds were available to assist departments wishing to hire qualified individuals. As noted earlier in this paper, the number of women faculty at UW has significantly increased in the past twenty years – from 323 in 1978 to 480 in 1998.
In 1989, the university initiated a spousal hire program, which attempts to find employment for partners of new faculty members. Central funds to provide salary support for up to three years are available to departments willing to hire a spouse or partner. This is meant to encourage departments to hire an individual even if an opening is not immediately available. The funds may be used for faculty or staff positions on campus.
A formalized training program for department chairs and particularly new department chairs was adopted. Among topics covered are hiring, recruitment, mentoring, and guidance in the tenure process. Additional training on racial and sexual harassment issues is also provided to department heads. New faculty members are invited to attend an orientation on the tenure process at UW.
Eleven years ago, the university conducted a study of 215 probationary faculty hired 8-10 years earlier (Reed et al. 1988). The analysis probed at what rate faculty left before tenure, why they left, whether they found the tenure process fair and whether there were differences in responses for men and women.
The 1988 study reported that only one-fourth of probationary faculty had a formal mentor within their department. As a result, in 1992 the Faculty Senate adopted changes to the official Faculty Policies and Procedures to establish a formal mentoring program for all probationary faculty. Each department is required to have written guidelines for tenure and to establish a mentor or mentor committee for each probationary faculty. The mentor committee meets with the mentee on an annual or more frequent basis.
In addition, the study noted that many individuals had established informal mentors. About half of the women who found informal mentors chose other women as mentors. A Women Faculty Mentoring Program was established in 1989. Under the program, all tenure-track women faculty were given the opportunity to be matched with a woman tenured faculty member from a different department but in the same broad discipline area. In 1993 the program was formalized and given some central funds. Between 55 and 80 percent of women assistant professors have had mentors assigned through the program in recent years.
Faculty members in the 1988 study gave many different reasons for leaving. As noted above, about 60 percent of the men and half of the women left because they believed they would not get tenure. Of those who left voluntarily, the most common reasons cited by both men and women were better career opportunities and higher pay. To address this problem, in 1989-91 the state legislature granted additional funds to help UW "catch-up" to the average salary levels of peer institutions. In addition, internal reallocations through the Quality Reinvestment Plan in 1992-93 were used to help maintain competitive faculty salaries.
Several policies aimed at improving the overall climate for women were introduced in the past decade. Faculty members may now request an automatic extension of the tenure clock after the birth or adoption of a child. In 1990, an associate vice chancellor position was created with special responsibility for gender issues. A faculty and staff committee on women in the university was re-appointed after being dormant for several years. The committee surveyed women faculty and staff about what gender-related issues they would like to see addressed on campus. A 1992 faculty pay equity study concluded that women’s salaries were 3.8 percent lower than those of comparable men (Hyde and Jones 1992). About $860,000 in permanent salary adjustments were paid to women faculty as a result of the study.
VI. Conclusions
Based on this study, one can conclude that UW-Madison has made significant progress in increasing the tenure rate for women faculty. The proportion of tenure-track women who have been granted tenure has increased from 40 percent in 1978-82 to 58 percent in 1987-91. On average, women spend a longer time in probationary status prior to being granted tenure.
Tenure rates vary by discipline area. Only about half of the men and women in the social studies area achieve tenure, compared to about 70 percent in the physical sciences and over 60 percent in the humanities. Women in the biological sciences are less likely than men are to be promoted.
Once granted tenure, women and men faculty are equally likely to remain on the faculty for ten or more years. Faculty members hired with tenure are less likely to be retained than faculty promoted to tenure.
The number of new faculty hired has not kept pace with the number of faculty leaving in recent years. The total faculty has decreased by over 10 percent since its peak in the late 1980s. As a result, the mix of tenured and tenure-track faculty and of younger and older faculty is changing. This will have implications for future turnover patterns. The retirement rate is expected to creep up slowly, from about 28 percent in the past ten years to 31 percent in the next decade.
Numerous additional studies could further inform us about faculty turnover at UW Madison and elsewhere. This approach could be used to study whether there are differences in the retention of minority and non-minority faculty. An examination of promotion rates to full professor could also be undertaken. Other breakdowns of the data, such as comparing turnover rates by college, may be illuminating. Women are more likely to receive an extension to the total time in the probationary period. Further analysis of tenure clock extensions may shed light on the tenure rate for men and women faculty. Additional data collection to determine why faculty members leave would be valuable. As noted earlier, additional event history analyses, controlling for explanatory variables such as years since degree and discipline, are in progress.
[TOP/Table of Contents] [I. Introduction] [II. How much is "normal" turnover?] [III. Faculty Retention After Ten Years][ IV. Cohort Analysis] [V. Policies to Reduce Turnover] [VI. Conclusions] [ VII. References] [VII. Footnotes]
[List of Tables and Charts]
VII. References
Brown, Byron W. and Stephen A Woodbury. "Gender Differences in Faculty Turnover" Staff Working Paper 95-34. Upjohn (W.E) Institute for Employment Research, Kalamazoo MI, 1995.
Coalition on the Status of Women Faculty at Miami, "Tenure Rate by Gender: Assistant Professors hired in Tenure-Track Positions 1982-1990" Miami University of Ohio, Oxford OH 1997. [http://www.sba.muohio.edu/cswam/retention/hiring4c.html]
Cornell University Office of Institutional Research. Cornell University Fact Book 1998. Ithaca NY [http://www.ipr.cornell.edu/FactBook/]
Ehrenberg, Ronald, Hirschel Kasper, and Daniel Rees. "Faculty Turnover at American Colleges and Universities: Analyses of AAUP Data", Economics of Education Review, vol. 10, no.2, pp99-110, 1991.
Eimers, Mardy T. "Exploring Faculty Career Progression: A Retention and Tenure Perspective" AIR national conference, Boston MA, May 1995.
Harrigan, Margaret N. Personal correspondence with staff at offices of institutional research at other research universities, Spring 1998.
Honeyman, David S. Jr. and Susan Robinson Summers. "Faculty Turnover: An Analysis by Rank, Gender, Ethnicity, and Reason" Proceedings of the 18th National Conference on Successful College Teaching, Orlando FL, February 1994.
Hyde, Janet S. and Donna M. Jones. "Gender Equity Study of Faculty Pay: University of Wisconsin - Madison" Madison WI June 1992.
Kingsbury Jones, Laura, Stephen A. Hoenack, and Mustapha Hammida. "Career Development of Tenure-Track Assistant Professors", Thought & Action, Winter 1994, vol. 9, no.2, pp.147-172.
Reed, Laurie, with Robin Douthitt, Bonnie Ortiz, and Diane Rausch. "Gender Differences in Faculty Retention at the University of Wisconsin – Madison" Madison WI, July 1988.
Schwab, Donald P. "Contextual Variables in Employee Performance-Turnover Relationships",
Academy of Management Journal, December 1991, vol. 34, No. 4, pp.966-975.
Singer, Judith D. and John B. Willett. "Modeling the Days of Our Lives: Using Survival Analysis when Designing and Analyzing Longitudinal Studies of During and Timing of Events", Psychological Bulletin, 1991, vol. 110, no. 2, pp.268-290.
South Dakota Board of Regents. Fact Book Fiscal 1997. Pierre SD 1997. [http://www.ris.sdbor.edu/publications/1997factbook/ ]
. Fact Book Fiscal 1998. Pierre SD 1998. [http://www.ris.sdbor.edu/publications/1998factbook/ ]
Tamada, Mike and Claudia Inman. "Survival Analysis of Faculty Retention Data: How Long Do They Stay?" AIR national conference Orlando FL, May 1997.
Tolbert, Pamela S., Tal Simons, Alice Andrews, and Jaehoon Rhee. "The Effects of Gender Composition in Academic Departments on Faculty Turnover" Industrial & Labor Relations Review, April 1995, vol. 48, no. 3, pp. 562-580.
University of Alaska Statewide Office of Institutional Research. "UA in Review 1997" Fairbanks AK, March 1997. [http://info.alaska.edu/UA/OIR/uair97/]
Yaffee, Robert A. and James T. Austin. "Discrete Time Event History Models for Higher Education Research" presented at the New York University Graduate Sociology Methodology Workshop on March 8, 1995.
VIII. Footnotes
1. The database I used includes a code for the reason an individual left employment, including categories such as resignation, non-renewal, retirement, dismissal, and death. However, an individual who is reasonably certain he will not get tenure will often resign before the formal tenure process takes place. This individual will therefore be coded as a resignation rather than non-renewal. For this reason, I have not distinguished between voluntary and involuntary leaving. Additionally, I have found a number of individuals in the data base who are classified as resigned although they worked at the university for over 30 years and were over 70 years old when they left. I have therefore treated all individuals who were 62 years of age or older when they left the university as if they retired, regardless of how they are actually coded on the database. (Return to footnote 1 position in text.)
2. Within the Medical School, faculty in clinical departments may choose either of two tracks: tenure or Clinical Health Sciences (CHS) track. CHS faculty have significant instructional and clinical practice responsibilities but are generally less involved in research. They have multi-year contracts but are not eligible for tenure. Faculty who switched from CHS to tenure track or vice versa (55 men and 10 women) are excluded from this analysis. (Return to footnote 2 position in text.)
3. As is common at most universities, UW-Madison follows the American Association of University Professors (AAUP) guidelines for the tenure process. Faculty governance documents establish a maximum probationary period of seven years of full-time service in the university in the ranks of instructor or assistant professor. The maximum probationary period may be decreased by up to three years for prior service at another university. Extension to the seven-year period may be granted for reasons such as part-time status, leave of absence, illness, childbirth or adoption, significant elderly or dependent care responsibilities, etc. (Return to footnote 3 position in text.)
4. The university has made several changes to its rules regarding extensions to the tenure clock during the period analyzed. Since 1994, any faculty member who requests it is automatically granted a one-year extension to the probationary period within one year of the birth or adoption of a child. (Prior to 1994, such extensions were approved on an individual basis.) In recent years, over half of all extensions to the tenure clock have been granted for reasons of childbirth or adoption. Of the current faculty, approximately 40 women and 24 men were granted one or more tenure clock extensions for childbirth or adoption. About half of these individuals have since been promoted to tenure. (Return to footnote 4 position in text.)
