by Peter Einaudi, Ruth Heuer, and Patricia Green[1]

The total number of postdoctoral appointees (postdocs) reported in the 2010 Survey of Graduate Students and Postdoctorates in Science and Engineering (GSS) grew to 63,415 in 2010, an increase of 10% over the 2009 total and 25% over the 2007 total. These 1- and 3-year growth rates are the highest in the history of the GSS and likely reflect improved reporting, as well as the continued expansion of postdoc employment in academia.

This InfoBrief assesses the impact of methodological changes to the GSS to reduce known reporting problems on the postdoc counts reported. Because the methodological changes began in 2007, analyses gauging the impact of the changes are based on the growth since 2006. As seen in figure 1, the number of postdocs reported in the GSS has climbed steadily over the duration of the survey, with a marked increase from 2007 through 2010. The results of this examination suggest that the 2010 postdoc data are the most accurate and comprehensive to date and that aggregate trends by discipline and demographics were largely unaffected by recent changes in reporting.

  Figure 1 Source Data: Excel file

Key Trends: 1979–2010

Postdoc Employment by Discipline

The GSS has collected data on postdocs annually since 1979. These data are widely used to estimate the trends seen in postdoc employment over the past 30 years. The dominant trend seen over this period has been the expansion of postdoc employment within and across disciplines (table 1). Because of the extra variability that may have resulted from the methodological changes made in the 2007 through 2010 GSS, all growth rate calculations comparing pre- and post-2007 counts are rounded to the nearest 5%. (See "Data Source and Limitations" for more information.)

The GSS data show that the number of postdocs increased across all science, engineering, and health (SEH) fields, with most fields experiencing substantial growth in each decade since 1980. Over the past decade (2000–10), postdoc employment in engineering had the fastest growth, with 8 of 14 engineering fields more than doubling the number of postdocs employed within them. Among the sciences, only computer sciences had a similar rate of growth.

As in 1979, the most common fields for postdocs in 2010 were the biological sciences, clinical medicine, and physical sciences fields. Even though the number of postdocs employed in the physical sciences has nearly doubled since 1979 (figure 2), the proportion of SEH postdocs in the physical sciences has declined substantially over time, dropping from 22.4% of all SEH postdocs in 1979 to 12.1% in 2010. By contrast, the proportion of SEH postdocs in engineering increased substantially over time, increasing from 5.9% of all postdocs in 1979 to 11.0% in 2010.

  Figure 2 Source Data: Excel file

As part of the 2007 survey redesign, five new fields (communication, family and consumer sciences/human sciences, multidisciplinary/interdisciplinary studies, neuroscience, and architecture) were added to the list of GSS eligible fields.[2] As seen in table 1, the number of postdocs in each of these fields has grown substantially since the fields were introduced to the GSS in 2007, especially in neuroscience and multidisciplinary/interdisciplinary studies. However, as shown in figure 2, these five new fields account for a small proportion of all postdocs. Because many of these postdocs were likely to have been reported under other SEH fields before the expansion of the GSS eligible fields list, the impact of these additional fields had limited effect on the overall increase in postdocs. Of the 1,673 postdocs reported in these new fields in 2010, a total of 913 were in units that existed under different GSS fields in 2006. Therefore, adding these five new fields accounted for an additional 760 postdocs, approximately 5% of the overall increase of approximately 14,100 in postdoc counts from 2006 to 2010.

Postdoc Employment by Demographic Characteristics

Women constituted 38.1% of the postdoc workforce in 2010, up from 18.5% in 1979 (table 2). Following 2 decades of growth, the proportion of foreign postdocs holding temporary visas leveled off in the early 2000s and declined from 2007 to 2009.

In 2010 the postdoc section was expanded to include data on the race and ethnicity of U.S. citizens and permanent residents. This key change was part of a larger expansion of the postdoc data collection designed to meet the analytical needs of the postdoc community, and it is discussed in greater detail below (see "Methodological Changes: 2007–10"). The 2010 GSS postdoc data indicate that blacks or African Americans and Hispanics or Latinos are significantly underrepresented within the postdoc ranks (table 2). Among all U.S. citizen and permanent resident postdocs in 2010, only 3.0% were black or African American and 3.9% were Hispanic or Latino. These figures are less than half of the percentages seen among graduate students enrolled in SEH fields in 2010 (8.0% and 7.3%, respectively) and even lower than the percentages found in the adult U.S. citizen population aged 21–45 (13.8% and 11.9%, respectively).[3]

Units Reporting Postdocs

A final indicator of the increasing prevalence of postdocs in SEH disciplines is the growth of the number of institutions and units (academic departments, programs, research centers, or health care facilities) reporting one or more postdocs. From 1979 through 2006, the number of institutions in the GSS that reported postdocs was relatively stable, whereas the number of units within institutions reporting postdocs grew steadily (table 3).[4]

As shown in table 3, the number and proportion of units reporting postdocs has increased almost every year since 1979. These increases serve as key indicators of the expansion of postdoc employment in the U.S. academic sector. However, the large increases in the number of units reporting postdocs since 2007 are due at least in part to methodological changes.

Methodological Changes: 2007–10

The recent rates of growth in institutions and units reporting postdocs and the count of postdocs were influenced, in part, by methodological changes between 2007 and 2010. In addition to the expansion of the GSS eligible fields discussed above, the National Science Foundation (NSF) and the National Institutes of Health (NIH) authorized four key changes in the GSS data collection designed to improve postdoc reporting (year of implementation in parentheses):

• Replacing the term "department" with "organizational unit" to capture nontraditional department-like places where postdocs work (2007)
• Encouraging schools to designate a coordinator specifically for reporting postdoc data (2008–10)
• Expanding the postdoc data items to collect more detailed counts by ethnicity, race, source and mechanism of support, and type and origin of doctoral degree (2010)
• Sending letters to institution presidents with a formal request to designate a Postdoc Coordinator and to increase institutions' participation in the postdoc data improvement efforts (2010)

Change in Terminology

The change in terminology from "department" to "organizational unit" was implemented in 2007 to explicitly include nontraditional departments and places were postdocs are employed, such as laboratories, centers, and other affiliated research units. This change was made because postdocs in these units may have been omitted in prior cycles because of the survey's focus on academic departments enrolling graduate students.

The increased reporting of postdocs in nondegree granting units (e.g., research centers and laboratories) in the GSS institutions is a key driver of the overall increase in postdoc counts from 2007 to 2010 (figure 3). After being relatively stable for over 20 years, the number of nondegree granting units reporting postdocs increased substantially between 2007 and 2010. Of the 21,393 postdocs reported in nondegree granting units in 2010, a total of 6,105 were in units added since 2006, representing 43.4% of the overall 14,072 increase in postdoc counts from 2006 to 2010.

  Figure 3 Source Data: Excel file

Postdoc Coordinator and Postdoc Data Expansion

Over the past decade, research sponsored by NSF and NIH consistently suggested that postdocs were being underreported in the GSS and other studies and that more detailed information, similar to what GSS collects on the graduate students, was needed to better understand the postdoc labor market.[5] Among the factors leading to difficulties in accurate reporting of postdoc data were institutional differences in how postdocs were defined and tracked and the lack of centralized recordkeeping systems for postdocs.

To improve the access and accuracy of the GSS postdoc data, NSF has been working with GSS institutions to identify individuals best qualified to provide their institution's postdoc data. Until 2008, schools typically appointed a School Coordinator (SC) to be responsible for reporting both student and postdoc data. To improve reporting, NSF was interested in the efficacy of appointing two coordinators: a Student Coordinator (StC) to report the graduate enrollment data and a Postdoc Coordinator (PC) to report information on postdocs and other nonfaculty doctorate researchers. In 2008, a total of 19 PCs were identified and given the ability to report the postdoc data for their schools and units independently from the graduate student data collection. In 2009, a Postdoc Pilot Study was conducted to determine (1) whether schools could provide detailed data on their postdocs' race and ethnicity, source of financial support, and type of doctorate degrees held and (2) whether having a separate PC would improve postdoc reporting. The results of the study, which included 48 small schools and 20 larger schools, confirmed the viability of both.[6]

A final objective of the 2009 Postdoc Pilot Study was to identify the common characteristics of a postdoc position across institutions. Respondents were asked to indicate which common characteristics of postdoc positions were applicable to postdocs at their institution. The following characteristics were reported by more than 80% of responding institutions:

• Requires a PhD or an equivalent doctorate degree
• Provides training in research
• Is intended to advance professional skills
• Requires that the postdoc work be under the direction of a senior scholar
• Is intended to prepare the postdoc for an independent career in research

These characteristics matched well with the GSS definition, which focuses on recent doctorate recipients with limited-term appointments primarily intended for training under the supervision of a senior scholar. The results also confirmed the substantial variation across institutions and identified the need to collect this information across all GSS institutions.

Because of the success of the Postdoc Pilot Study, the postdoc section of the 2010 GSS survey was substantially expanded to align with the graduate student items and to include questions concerning the criteria used to define postdocs. All schools were also requested by NSF to designate a separate PC as needed.

Letter to Institution Presidents

To highlight these survey changes and to increase institutions' participation in the postdoc data improvement efforts, a letter was sent from NSF and NIH in 2010 to the president of each institution emphasizing the importance of postdoc data in GSS and providing summary information on the postdoc counts reported by their institution in the 2008 and 2009 GSS. If these data were not accurate, the letter asked the president to designate a PC.

In 2010, a total of 125 schools opted to split the GSS data collection responsibilities between a StC and a PC. Of the 567 schools that retained a single SC, 104 schools replaced the 2009 SC with a new SC. table 4 and 5 show the results of these changes, by looking at changes in postdoc reporting by coordinator type.

Several key distinctions were found between schools that used PCs and those that did not (table 4). First, schools where postdocs were spread across many units were more likely to use a PC than were schools where the postdocs were concentrated within a few units. Postdocs were reported in a higher proportion of units in schools with a PC than in schools that had a single SC. For example, 68.1% of the units in schools with a PC in 2008 had postdocs as compared to 36.1% of the units in schools with an SC.

Second, new PCs were much more likely to add new units than were SCs or PCs in schools that had designated a PC in a prior cycle; new units with postdocs represented 11.9%, 11.6%, and 8.8% of all units in schools with new PCs in 2008, 2009, and 2010, respectively. As shown in table 5, of the 63,415 postdocs reported in 2010, a total of 2,358 were in units added by PCs in the prior three cycles, representing 16.8% of the overall 14,072 increase in postdoc counts from 2006 to 2010 (table 3).

Finally, as shown in table 4, in both 2009 and 2010 no new postdoc units were added within schools that currently had a PC and had used a PC in a prior cycle. Together, these differences provide substantial evidence that new PCs improved survey coverage by adding previously unreported, but eligible, units. These findings also suggest that improvements in coverage are a one-time phenomenon occurring the year a PC is added.

Table 5 provides additional evidence that the addition of a PC improved postdoc reporting and that this effect was primarily felt in the year the PC was added. As seen in the percentage change for the prior year, new PCs increased overall postdoc reporting in their schools by 17.0% in 2008, 46.6% in 2009, and 15.6% in 2010. Each of these increases is significantly higher than was typical prior to 2008 and greater than their peers (SC and PC in a school that had used a PC in a prior year) in each of these years. Although some of the increases in postdoc counts reported by new PCs were associated with newly added units, new PCs also increased the number of postdocs reported within extant units at a higher rate than SCs or PCs in schools that had designated a PC in a prior cycle.

This finding demonstrates that PCs also improved survey accuracy by adding postdoc counts to units where prior SCs could not. New PCs increased postdoc reporting within extant units by 10.5% in 2008, 16.6% in 2009, and 5.8% in 2010. Finally, increases in postdoc counts were lowest in 2009 and 2010 among schools that had previously used a PC, underscoring the notion that the primary impact of adding a PC is achieved the year the PC is added.

Overall Impact of Methodological Improvements on Postdoc Counts

The overall impact of these methodological changes on the postdoc data is difficult to estimate. Two different ways of estimating the impact can be examined.

First, there is the directly observed impact of the methodological changes. In the 2010 data collection, 7,160 postdocs were reported in 879 units that meet at least one of the following criteria:

• The unit was added in one of the new fields added to the GSS code list in 2007.
• The unit was a nondegree granting unit added since 2006.
• The unit was added by a PC.

This count represents 50.9% of the overall change in the number of postdocs reported from 2006 to 2010.

Another method of estimating the impact would be to assume the growth of postdoc employment would have remained steady over this period. From 1979 through 2006, postdoc employment grew by an average of 1,157 postdocs per year. Based on this average growth, expected growth from 2006 through 2010 is estimated at 4,628 and growth due to the methodological improvement is estimated at 9,444, or 67.1% of the observed growth.

Data Source and Limitations

Cosponsored by NSF and NIH, the GSS is an annual survey that provides data on the number and characteristics of graduate students, postdocs, and other doctorate-holding nonfaculty researchers in science, engineering, or selected health fields in all U.S. academic institutions granting doctorate and research-based master's degrees in these fields.

This InfoBrief is based on data published in the 2010 GSS. The 2010 GSS collected data on the graduate students and postdocs from 13,711 organizational units (departments, programs, affiliated research centers, and health care facilities) at 574 institutions of higher education and their affiliates in the United States, Puerto Rico, and Guam. The institutional response rate was 99.3%. An overview of the survey can be found at http://www.nsf.gov/statistics/srvygradpostdoc/.

In 2007, eligible fields were reclassified, newly eligible fields were added, and the survey was redesigned to improve coverage and coding of eligible units. "2007new" presents data as collected in 2007, and "2007old" presents data as they would have been collected in 2006. Due to methodological changes in 2007, the data collected from 2007 through 2011 are not strictly comparable to those collected prior to 2007. As a result, care should be used when assessing trends within the GSS data. Ten-year trends reported in the tables are labeled "% change 2002–11." Note that these percentages are rounded to the nearest 5% and counts are rounded to the nearest 100 to reflect the extra variability in the estimate that may have resulted from the methodological changes that occurred in 2007. Please see appendix A, "Technical Notes," in Graduate Students and Postdoctorates in Science and Engineering: Fall 2007 (NSF 10-307) for a more detailed discussion of these changes.

Reporting of ethnicity and race in 2008–11 has been affected by changes in reporting of ethnicity and race in the Integrated Postsecondary Education Data System (IPEDS). Starting in 2008, IPEDS respondents were asked to use a new classification that included a category for two or more races (see http://nces.ed.gov/ipeds/reic/resource.asp) and separate reporting of Native Hawaiians and Other Pacific Islanders from Asians. The new classification was optional in 2008 and 2009 IPEDS but mandatory in 2010 and may have contributed to a significant increase in reporting of "Not Hispanic or Latino, More than one race" within the GSS data.

Each of the major methodological changes in GSS resulted in improved reporting of postdoc data. The expansion of the GSS code list in 2007, the change in focus from degree-granting graduate programs to eligible units regardless of degree-granting status, and the appointment of separate PCs improved coverage of postdocs in SEH fields. The expansion of the postdoc data items and appointment of more knowledgeable PCs improved postdoc data reporting within extant units.

In addition to improving the reporting of overall postdoc counts, the 2010 GSS provided more detailed information on postdocs, including the ethnicity and race of U.S. citizens and permanent residents; federal and nonfederal sources of financial support and support mechanism; type of doctoral degree, such as MD, PhD, or joint MD and PhD; and origin of doctoral degree (U.S. or foreign country).[7]

Notes

[1] Peter Einaudi and Ruth Heuer are research analysts and Patricia Green is a survey director at RTI International. For further information, contact Kelly H. Kang, Human Resources Statistics Program, National Center for Science and Engineering Statistics, National Science Foundation, 4201 Wilson Boulevard, Suite 965, Arlington, VA 22230 (kkang@nsf.gov; 703-292-7796).

[2] A complete list of GSS fields can be found in appendix B of Graduate Students and Postdoctorates in Science and Engineering: Fall 2010 (http://www.nsf.gov/statistics/nsf13314/).

[3] Einaudi P. 2011. Two Decades of Increasing Diversity More than Doubled the Number of Minority Graduate Students in Science and Engineering. InfoBrief NSF 11-319. Arlington, VA: National Science Foundation, National Center for Science and Engineering Statistics. Available at http://www.nsf.gov/statistics/infbrief/nsf11319/.

[4] From 1984 to 1987, the number of GSS institutions dropped substantially as master's-granting institutions were subsampled and counts for the nonsampled institutions were imputed within two placeholder institutions labeled as Unsampled Public Master's and Unsampled Private Master's. As a result, the proportion of institutions reporting postdocs prior to 1984 and after 1987 is not comparable.

[5] McFarland E, Einaudi P, Cook S, Richards A, Roe D, Zwieg E, Green P. 2010. GSS Recordkeeping Study. Report to the National Science Foundation. Research Triangle Park, NC: RTI International.

[6] McFarland E, Steele B, Zwieg E, Green P. 2011. GSS Postdoc Pilot Study. Report to the National Science Foundation. Research Triangle Park, NC: RTI International.

[7] National Science Foundation, National Center for Science and Engineering Statistics. 2013. Graduate Students and Postdoctorates in Science and Engineering: Fall 2010. Detailed Statistical Tables NSF 13-314. Arlington, VA. Available at http://www.nsf.gov/statistics/nsf13314/.