Summary of Findings
During the spring and summer of 2004, the Cornell Higher Education Research Institute (CHERI) conducted a survey of public universities to document the trends over time in graduate teaching assistants’ (hereafter referred to as TAs) compensation nationwide. Specifically, the survey requested information on the minimum academic year stipend paid to full-time TAs and whether the stipend increases with experience. The survey also asked about the dollar value of any tuition remission received by full-time TAs, by residency status. Lastly, the survey requested information on university-provided health insurance for full-time TAs and the proportion (or dollar value) of the university contributions made on behalf of these assistants beyond the contributions made for full-time graduate students not employed as full-time teaching or research assistants. The survey asked for this information for academic years 1989-90 to 2003-04, at two year intervals, as well as for three more distant academic years: 1975-76, 1980-81 and 1985-86.
The survey was distributed to 178 four-year public universities that conferred a significant number of doctoral degrees and were located in the fifty states. As a result, nearly a majority of the institutions surveyed were Research I (33%) or Research II (16%) institutions as classified by the Carnegie Foundation in their 1994 classification of institutions of higher education. Of the 178 surveyed institutions, 111 responded, generating an overall response rate of 62.4%. Table_1.pdf (PDF, 41 KB) presents the response rate by Carnegie classification.The response rate was highest for Research I universities (76%) and lowest for Doctoral I institutions (46%). Since there are so few Comprehensive II and Engineering Institutions in the sample, institutions in these two categories are combined with those in the Comprehensive I category for the purpose of reporting results below.
The survey began by asking for information on the minimum academic year stipends for full-time TAs.Table_2.pdf (PDF, 44 KB) presents summary information about the percentile distributions of stipends for the entire sample. All the stipend levels have been adjusted to 2004 dollars. With the exception of the first academic year for which information was provided, 1975-76, the TA stipend monotonically increases in real terms with time; the decline that took place between the 1975-76 and 1985-86 academic years may be due either to the changing sample size over time or to price increases outpacing increases in stipends during this period. The median stipend reaches its peak of $9,916 in the 2003-04 academic year, representing a 34 percent increase in real terms since 1980-81. The largest percentage point increase occurred between the most current academic years 2001-02 and 2003-04 during which the median stipend increased 9 percent in real terms.
Table_2.pdf (PDF, 44 KB) also presents the data on the percentage of institutions that pay higher stipends to TAs with more experience. The results show that the proportion of institutions that base stipends, in part, on experience is remarkably consistent over time, at approximately two-thirds. The percentage of institutions that pay such experience differentials is highest in 1985-86 at 72 percent and lowest in 1999-00 at 67 percent.
Table_2a.pdf (PDF, 45 KB) provides basic data on the percentile distribution of stipends, restricting attention to the 46 institutions that provided stipend data for each academic year for which information was requested between 1989-90 and 2003-04. The findings are generally similar to those from Table 2a in that the stipends have been on a near-monotonic rise in real terms during the fifteen-year period. However, between 1993-94 and 1995-96, the median stipend declined in real terms suggesting that stipends may not have been keeping pace with inflation during that time period. The proportion of institutions that paid higher stipends to TAs with more experience is higher in this sub-sample, averaging 75 percent across the 15-year period compared to 68% in the entire sample. The primary reason for this difference is the high representation of R1 institutions in the sub-sample and the fact that R1 institutions are much more likely to pay experience differentials than non-R1 institutions.
Table_3.pdf (PDF, 47 KB) provides summary information on TA stipends, disaggregated by the 1994 Carnegie classification. In every academic year for which data is available, RI institutions have the highest median stipends, roughly 24 percent higher than the median for all institutions. With the exception of two years, all of the other Carnegie classifications have average stipends lower than the average for all institutions. Another consistent trend is that since 1985-86, DI institutions have paid less, based on the median stipend, than those institutions classified as DII. This may be a consequence of location differences between the two groups; in particular, DII schools that responded to the survey are more likely to be located in urban areas than the DI institutions that responded. Similarly, since 1985-86 the comprehensive institutions in our sample also have paid higher stipends than the DI institutions (and for some years, more than the DII institutions) for likely the same reason; a higher proportion institutions are located in urban settings of the comprehensive where the cost of living is higher.
Table_4a.pdf (PDF, 45 KB) and Table_4b.pdf (PDF, 45 KB) provide information on the tuition remissions provided for in-state and out-of-state graduate students who work as TAs, respectively. Specifically, the tables provide the mean percentage tuition remission and the percentages of institutions that remit 100 percent and that remit more than 50 percent of tuition costs. Looking at in-state tuition for all institutions first, the mean percentage of TA tuition remission has been climbing steadily during the period; rising from 74 percent in the 1975-96 academic year to 86 percent in 2003-04. The percent of respondent institutions that pay 100 percent of TA tuition was around two-thirds during the entire period. On the other hand, the proportion of respondents that pay more than 50 percent of TA tuition has been steadily increasing, with over eighty percent of sample institutions paying at least 50 percent of tuition since 1993-94.
Disaggregated by 1994 Carnegie classification, a somewhat surprising pattern emerges. The DII institutions have the highest mean percentage tuition remission in all periods except for the very earliest, when the sample sizes were extremely small. Not surprising, these institutions are also typically the most likely to provide 100 percent of tuition or to provide more than 50% in tuition remissions. Unlike the situation with stipends, in which RI institutions brought up the average for all institutions, the mean percentage tuition remission for RI institutions is typically very near the mean percentage remission rate for all institutions and is, in some years, below the all-institution rate.
An even more surprising result emerges in examining the tuition remissions for out-of-state TAs presented in Table_4b.pdf (PDF, 45 KB). With the exception of two most recent years, the mean percentage remission is greater for out-of-state TAs than in-state TAs. In addition, while the mean percent remission has risen for in-state TAs over the period, the mean percent remission for out-of-state TAs has been generally stable around 86 percent. However, the patterns observed for in-state remissions by Carnegie classification generally hold for out-of-state remissions as well. For example, D1 institutions again provide the highest mean percentage tuition remissions and are generally the most likely to provide 100 percent tuition remissions and to provide 50 percent or more in tuition remission during the period.
Table_5.pdf (PDF, 43 KB) presents the proportions of respondents that indicated that the university offsets the cost of health insurance by more than they provided to graduate students who are not employed as teaching or research assistants. Examining the broad trends for all respondents, the proportion of institutions that contribute to teaching assistant health insurance costs, at least in part, increased monotonically with time. In the most recent survey year, 56.4 percent of respondents reported paying some portion of health insurance costs for teaching assistants compared to less than 15 percent in 1985-86. Examining the findings by 1994 Carnegie classification, yields generally similar results. RI institutions are the most likely to contribute to health insurance costs, with a majority of respondents having made such payments since the 1993-94 academic year. In the most recent year, 2003-04, 80% of RI institutions contributed to the health insurance costs of teaching assistants. Similar to the findings for stipends, the proportion of Comprehensive institutions that made health insurance contributions exceeded that of DI institutions in all survey years; further, the percentage of DII institutions also exceeded that of DI institutions in recent years. These findings again may result from the more urban locations of the comprehensive and DII institutions relative to those classified as DI.
Due to the small sample sizes, generalizing our findings to all public higher education institutions, and particularly those by Carnegie classification, should be done with caution. Nevertheless, our findings suggest that there has been a significant increase in the compensation packages – including stipend, tuition remission and health insurance -- of teaching assistants in public academic institutions that confer doctoral degrees. Further, the improvement of teaching assistant compensation has taken place across the board at all Carnegie classification levels. It is the case, however, that the most prestigious public research institutions, those classified as Research I by the 1994 Carnegie classification scheme, pay the highest stipends and are the most likely to help defray the cost of health insurance. On the other hand, RI institutions do not provide the largest tuition remissions in percentage terms.
 Medical and/or health science institutions were excluded from the surveyed sample since these schools and their programs often operate distinctively from commonly understood graduate studies programs.
 All of the institutions have some doctoral programs even though the Carnegie Foundation classified some of them as comprehensive or specialized engineering schools in 1994.
 In order to maintain confidentiality, the names of the surveyed institutions as well as those of the respondents are not reported here.
 One of the primary components of this decline was a significant drop in the reported stipend levels at two of the institutions. Since there was no significant change in the inflation rate during this period, the price index itself does not explain this decline.
 This finding leads us to worry that some respondents may not have fully understood our questions and hence our findings on tuition remission should be considered very tentative.