Publishing In The Postdoc Pipeline Rev 010208
Upcoming SlideShare
Loading in...5
×
 

Publishing In The Postdoc Pipeline Rev 010208

on

  • 1,918 views

This draft manuscript was never able to be published because the statistics required to pass seemed a bit too daunting, so now I allow it be viewed publicly. This is an analysis I did to see if I can ...

This draft manuscript was never able to be published because the statistics required to pass seemed a bit too daunting, so now I allow it be viewed publicly. This is an analysis I did to see if I can calculate the first-author "publication rate" of postdocs from self-disclosed responses in the 2004 Sigma Xi Postdoctoral Survey. I also looked at variations in the way a publication rate can be viewed based on various demographic and training outcomes.

If you have comments in how I can improve or publicize the results of this document, please let me know.

Statistics

Views

Total Views
1,918
Views on SlideShare
1,918
Embed Views
0

Actions

Likes
0
Downloads
10
Comments
0

0 Embeds 0

No embeds

Accessibility

Upload Details

Uploaded via as Adobe PDF

Usage Rights

CC Attribution License

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

    Publishing In The Postdoc Pipeline Rev 010208 Publishing In The Postdoc Pipeline Rev 010208 Document Transcript

    • Publish or Perish: Peer-Reviewed Publication Rates of Postdocs as a Metric for Productivity and Effective Scientific Training CONFIDENTIAL WORKING DRAFT ONLY First draft (December 29, 2006) by Emil Thomas Chuck, Ph.D. Health Professions Advising Office Student Academic Affairs & Advising George Mason University 4400 University Drive (Mail Stop 2C4) Fairfax, Virginia 22030 Co-chair, Diversity Committee National Postdoctoral Association Document revised 1/10/2008 9:12:11 AM Submit for Sigma Xi or Journal of Women and Minorities in Science and Engineering
    • Abstract The production of peer-reviewed first-author papers is a significant metric for an individual postdoc’s future career success in academe. To assess the productivity of the American postdoctoral training system, data from the 2004 Sigma Xi Postdoctoral Survey were analyzed. 38% of the survey respondents had not published a first-author paper, and the average number of first-author publications was 2.3 among those who had published. Published postdocs on temporary visa generally claim more first-author papers (2.5) compared to native-born postdocs (2.0). Among American-citizen published postdocs, Asians and Hispanics report more first-author publications compared to Caucasians and African-Americans. Looking at gender effects, published postdoctoral women claim fewer first-author publications compared to men, and published men which children claim much more first-author papers compared to men without children or all women. Finally, our data suggest that formalized mentoring structures are positively correlated with a higher number of first-author publications.
    • Introduction Much attention has been made about the lack of women in scientific leadership positions, whether it is in academia, government, or industry. While a lot of recent progress in placing women scientists in prominent administrative positions has garnered some attention, the lack of overall progress in recruitment, retention, and promotion throughout the scientific workforce pipeline has been frustrating. Many studies show that psychological metrics improve for women and minorities in scientific fields in environments that value their demographic and personal identities, allow for flexibility in work-life balance, and address issues related to implicit biases. However, there is very little information that looks at how specific interventions could make women and minority scientists more viable to hire as principal investigators. Indeed, one of the key measures for measuring an individual scientist’s professional merit is the number of peer-reviewed publications, and the ultimate standard of a productive postdoctoral fellowship has generally been accepted as the first-author peer-reviewed publication. Many studies have shown that women and minorities tend to publish fewer papers as faculty members, which jeopardize their standing for promotion. The AAMC disclosed that women may produce “better” papers with higher impact per publication. However, no studies have looked at the productivity of postdoctoral scientists who could be potential candidates for hire. In 2004, the Sigma Xi Postdoctoral Survey became the first national/international survey that looked into the state of postdoctoral training in the United States. An analysis of the survey revealed significant differences in the postdoctoral experience of women and underrepresented minorities compared to the overall cohort. Women claimed fewer sole- author, first-author, and non-first author peer-reviewed articles than men, and women accommodated more of their schedule to adjust for better work-life balance. Women and minorities reported a lower quality of mentoring and guidance in their postdoctoral training compared to the overall cohort. However, our previous reports did not look at significant details in these findings based on research discipline, time in postdoctoral training, or demographic. The objective of this study is to see whether there were differences in publication rate of first-author peer-reviewed publications among women and minority postdoc respondents in the survey. In order to do this, a comprehensive analysis of overall postdoctoral publication had to be undertaken, with many of the results disclosed here in this report. The analysis is meant to serve as a snapshot of the overall workforce pipeline and their ability for the population to produce first-author papers. This allows for this analysis to control for the length of time taken for each postdoctoral respondent. To our knowledge, this study is the first to take a look at the productivity of postdoctoral training as evidenced by self-reported creation of peer-review articles. This paper serves to address significant questions raised about the productivity of the American postdoctoral training system and identify the major drivers of innovation in the system.
    • Selection of data While many core questions are asked to all postdocs in the survey, many of the questions focusing on publications, postdoctoral tenure, and type of institution were given selectively to scientific samples of the entire postdoctoral cohort. Statistical outliers included individual responses that claimed more than 10 publications within the first 2 postdoc years, and when appropriate, these outliers were removed for overall publication rate analysis. Sole author papers are not included. In our reporting, detailed prefer-no-answer and don’t know responses were excluded from this report but may be included in some aggregate “totals.” The methodology of data collecting for the 2004 Sigma Xi survey was previously published. SPSS 10.0 and Excel 2004 were used to manage the data from the 2004 survey to produce the reports shown in the data tables and charts.
    • Results How many papers are postdoctoral scholars producing? According to the survey data, approximately 8500 first-author publications were produced by 3715 postdocs (62%) in the respondent pool of 5960. While the calculated productivity of postdocs is 1.4 first-author papers per postdoc in the entire respondent pool, 2.3 first-author papers were actually produced per postdoc that published at least one first-author paper (to be termed as “published postdoc”). 35% of the respondents had two or more first-author publications. Within the respondent population, we looked at differences in the publication rate among published postdocs. Table 1 shows the results of this demographic analysis. When broken down by citizenship status, 1024 of 2966 postdocs on temporary visa (35%) claimed no first-author publications compared to 1021 of 2373 US native-born postdocs (43%), 97 of 220 US naturalized postdocs (44%), or 89 of 375 permanent resident postdocs (24%). 1210 men (35% of all male postdocs) and 1019 women (41% of all female postdocs) claimed zero first-author publications. Table 1. First-author publications per published postdoc by demographic and institution type First-author Cohort Total Published Publication Ratio Publications All postdocs 5979 3728 8551 2.29 US native 2373 1352 2668 1.97 US naturalized 220 123 245 1.99 Permanent Resident 375 286 875 3.06 Temporary Visa 2966 1942 4792 2.47 Male 3429 2219 5899 2.66 … US citizen or resident 1332 787 1778 2.26 … other citizenship 2097 1432 4121 2.88 Female 2498 1479 3166 2.14 … US citizen or resident 1356 758 1444 1.91 … other citizenship 1142 721 1722 2.39 Hispanic 116 64 154 2.41 Caucasian 2281 1368 2763 2.02 African-American 102 43 89 2.07 Native American 24 18 55 3.06 Hawaii/Pacific Islander 12 7 35 5.00 Asian-American 421 268 753 2.81 Government research 535 371 858 2.31 Medical institute 379 236 515 2.18 Private academic 2427 1446 3194 2.21 Public academic 1698 1089 2707 2.49 Research institution 896 559 1321 2.36 With children 1953 1365 3568 2.61 … Male with children 1239 876 2514 2.87 … Female with children 697 476 1044 2.19 No children 3905 2274 4815 2.12 … Male no children 2115 1290 2898 2.25 … Female no children 1758 968 1917 1.98
    • Since most of the postdocs that responded to the survey came from biomedical backgrounds, it interested us if there were differences in publication rate based on the research area. The tables below shows selected disciplines and their publication rates. In general biomedical science fields had a lower publication ratio compared to other science and engineering fields. Table 2. First author publication ratio by discipline Table 2a. Academic disciplines Discipline Postdocs Published postdocs First-author pubs Ratio Agricultural and Resource Economics 4 2 7 3.50 American Studies 5 0 0 --- Animal Sciences 50 38 77 2.03 Anthropology 19 7 14 2.00 Applied Mathematics 60 51 119 2.33 Astrophysics and Astronomy 62 53 185 3.49 Biochemistry, Biophysics, and Structural Biology 934 567 1355 2.39 Cell Biology 1010 604 1247 2.06 Chemistry 417 255 672 2.64 Comparative Literature 1 0 0 --- Computer Science 70 55 191 3.47 Developmental Biology 368 208 419 2.01 Earth Sciences 86 66 201 3.05 Ecology and Evolutionary Biology 164 124 341 2.75 Economics 14 7 10 1.43 English Language and Literature 3 0 0 --- Food Science and Engineering 8 6 23 3.83 Genetics, Genomics, and Bioinformatics 651 393 858 2.18 Geography and Regional Science 7 4 6 1.50 Global Cultural Studies 3 0 0 --- History 10 2 2 1.00 History of Art, Architecture, and Archaeology 6 2 2 1.00 Immunology and Infectious Diseases 511 315 644 2.04 Linguistics 3 2 2 1.00 Mathematics 20 14 42 3.00 Microbiology 344 207 487 2.35 Molecular Biology 1225 733 1610 2.20 Neuroscience and Neurobiology 588 384 835 2.17 Nutrition 42 30 63 2.10 Oceanography and Atmospheric Sciences and Meteorology 34 28 76 2.71 Pharmacology, Toxicology, and Environmental Health 134 79 223 2.82 Philosophy 2 0 0 --- Physics 251 173 663 3.83 Physiology 180 129 317 2.46 Plant Sciences 100 67 181 2.70 Political Science 6 1 1 1.00 Psychology 173 126 333 2.64 Sociology 22 11 25 2.27
    • Statistics and Probability 29 22 42 1.91 Table 2b. Clinical disciplines Clinical Area Postdocs Published postdocs First-author pubs Ratio Anesthesiology 14 8 18 2.25 Biostatistics and Clinical Trials 19 8 14 1.75 Cardiology 99 70 164 2.34 Colon and Rectal Surgery 3 2 2 1.00 Community and Environmental Medicine 3 1 1 1.00 Critical Care Medicine 13 8 10 1.25 Dental/Oral Surgery 3 2 4 2.00 Dentistry 8 3 25 8.33 Dermatology 15 9 26 2.89 Epidemiology 64 54 136 2.52 Family Medicine 1 0 0 --- Gastroenterology 31 15 46 3.07 General Surgery 5 2 4 2.00 Geriatric Medicine 7 3 9 3.00 Hematology 96 68 159 2.34 Immunology and Infectious Diseases 77 44 90 2.05 Internal Medicine 29 13 19 1.46 Medical Genetics 28 18 51 2.83 Medicine, General 41 26 50 1.92 Neonatal-Perinatal Medicine 6 4 6 1.50 Neurological Surgery 9 1 8 8.00 Neurology 49 37 83 2.24 Neurosciences 169 118 277 2.35 Nuclear Medicine 26 19 116 6.11 Nuclear Radiology 14 9 65 7.22 Nursing 5 4 10 2.50 Obstretrics and Gynecology 18 15 20 1.33 Oncology 265 158 354 2.24 Ophthamology 34 20 42 2.10 Optometry 2 1 6 6.00 Orthopedics/Orthopedic Surgery 17 12 31 2.58 Otorhinolaryngology 17 12 26 2.17 Pediatrics 61 27 54 2.00 Pharmacology 64 37 105 2.84 Physical and Rehabilitative Medicine 10 6 12 2.00 Plastic Surgery 4 1 2 2.00 Preventative Medicine 13 6 20 3.33 Psychiatry 75 50 116 2.32 Public Health 45 33 80 2.42 Pulmonary Medicine 38 18 26 1.44 Radiation Biology/Radiobiology 29 17 86 5.06
    • Thoracic Surgery 6 3 15 5.00 Urology 22 15 26 1.73 To determine whether publication productivity could be correlated to the source of funding for the postdoc, productivity of published postdocs was also analyzed with respect to the self-disclosed source of research funding (Table 3). These data also show that postdocs on biomedical science grants tend to claim fewer first-author publications compared to postdocs on other science grants, again emphasizing the disciplinary differences in the calculation of the overall publication rate. Overall, there is a higher ratio of first-author papers per published postdoc when the postdoc is financially supported by a supervising principal investigator compared to direct fellowship grant support. Table 3. First-author paper productivity based on research support. Type of grant Total Published PD Papers Ratio NIH Research grant 132 74 170 2.30 Prefer not to answer 12 7 38 5.43 NIH NRSA 280 180 407 2.26 NSF Research grant 17 13 49 3.77 NSF fellowship 22 17 71 4.18 DOE grant 3 2 28 14.00 Other US government grant 68 48 146 3.04 Non-US government grant 127 69 165 2.39 Foundation or non-profit source fellowship 487 289 591 2.04 Other 112 62 178 2.87 Postdoc source of funding Total Published PD Papers Ratio Grant to PI 2834 1832 4440 2.42 Grant to consortium 777 458 979 2.14 Grant to postdoc 1305 793 1708 2.15 Funds from institution/employer 654 403 886 2.20 Personal resources 23 11 46 4.18 Don't Know 237 132 382 2.89 Other 119 78 209 2.68 Prefer no answer 22 15 55 3.67 Not applicable/No answer 12 8 41 5.13
    • OVERALL PRODUCTION TRENDS IN FIRST-AUTHOR PEER REVIEW PUBLICATIONS Considering the significant number of postdocs who have not produced a first-author publication, we wondered how long it took for postdocs to produce their first and subsequent first-author publications. For the rest of this paper, our analysis considered the postdoctoral training time taken by the respondents at the time of the survey to calculate a publication rate metric for producing a first-author publication. In this section, we discover the self-reported number of postdocs without a first-author publication per year of postdoc, the number with at least one first-author publication, and the general publication rate among published postdocs per postdoc tenure year. Chart 1 shows the population of the 5782 respondents broken down by postdoctoral tenure. While the cohort contains postdocs with up to 10 years of experience, 83% of the respondents are in their first four years of postdoctoral training, and half of all respondents are in their first two years of postdoctoral training. Two populations are highlighted: postdocs who did not disclose publishing a collaborative first-author peer- reviewed article, and those postdocs that did. As the number of postdocs decreases with increasing tenure, the proportion of postdocs with first author publications increases. This trend is shown in Chart 2. To determine how quickly first-author publication were produced by postdocs, the number of first-author publications was asked of each of the published postdocs, and the average number of publications for each postdoc year was calculated (Chart 3). Published postdocs in their first or second year of tenure claimed an average of 1.9 first- author publications, though it is not clear whether the publications were related to delayed publication from doctoral studies or new publication related to their postdoctoral studies. Linear regression on these data suggests that the average publication rate was 0.51 first-author publications per postdoc year (R2 = 0.8451). The publication rate between postdoc years 2 and 5 was 0.26 first-author publications per published-postdoc year (R2 = 0.9725), 0.39 if one includes all unpublished postdocs (R2 = 0.9755).
    • Chart 1: First-author publications Chart 2: Percentage of postdocs with first-author publications 1600 1400 Unpublished 100% 1200 Published 80% 1000 Number of 60% 800 Number of post docs 600 Published postdocs 40% 400 Unpublished 200 20% 0 0% -1 -3 -5 -7 -9 -1 -3 -5 -7 -9 Y0 Y2 Y4 Y6 Y8 Y0 Y2 Y4 Y6 Y8 Postdoc tenure (years) Postdoc tenure (years) Chart 3: First-author paper publication Chart 4: Non-first-author paper rate publication rate 7 7 6 6 5 5 4 4 3 3 2 2 1 1 0 0 Po st d o c t enur e ( year s) Po st d o c t enur e ( year s) Chart 5: First-author publications: US native vs. Temp Visa Chart 6: Percentage cohort postdocs with first author 1400 publications 1 1200 US Native 0. 9 TV pub Temp Visa 0. 8 1000 US pub 0. 7 TV unpub 800 US unpub 0. 6 Number of Percentage of postdocs cohort 0. 5 600 0. 4 400 0. 3 0. 2 200 0. 1 0 0 -1 -3 -5 -7 -9 Y0 Y2 Y4 Y6 Y8 Postdoc tenure (years) Postdoc t enure (years)
    • A similar analysis was done for non-first-author peer-reviewed publications. The publication rate for published postdocs is shown in Chart 4. The proportion of postdocs with a non-first-author publication (58%) was similar with the proportion of all postdocs with first-author publications (62%). Overall, the publication rate was 0.51 non-first- author publications per postdoc year (R2 = 0.8311). Between postdoc years 2 and 5, the publication rate was 0.38 non-first-author publications per published-postdoc year (R2 = 0.9911), 0.50 if one includes all unpublished postdocs (R2 = 0.9874). Respondent data were also used to determine the median time it took for postdocs to produce first-author and non-first author peer-reviewed publications. Overall it took 16 months for a postdoc to produce his/her first first-author paper as a postdoc and 19 months for the first secondary-author paper. When trying to calculate the median time to produce subsequent publications, it appeared that the next first-author paper was produced when the postdoc had 42 months of training (26 months later), and a third paper by 66 months (24 months after that). In contrast, the next secondary-author papers appear to be produced at postdoctoral months 30 (11 months later) and 46 (16 months later). Publication Rates For US-born Native Postdocs and Postdocs On Temporary Visa We were interested in the differences in the publication rate of postdocs on temporary visa compared to native-born American citizens. The self-reported data show that overall, postdocs on temporary visa self-reported higher publication rates compared to US native colleagues. We focused on the first-author publication rate of postdocs with either a US Native (n=2109) or temporary visa citizenship status (n=2678). Chart 5 shows the population of published and published US native and temporary visa postdocs. While more postdocs are publishing papers with increasing postdoctoral tenure, the proportion of postdocs on temporary visa with first-author publications was larger than the proportion of US native postdocs. Overall, twice as many temporary-visa postdocs have first-author publications as there are US native postdocs. Chart 6 shows the calculated percentage of postdocs with first-author publications. By the fourth postdoctoral year, roughly 80% of all temporary visa postdocs had first-author publications while roughly 45% of similar- tenured US-native postdocs had first-author publications. The number of self-disclosed first-author publications was calculated per postdoctoral year (Chart 7). Most of the growth in the number of publications per published postdoc occurred among the more senior postdocs, with a dramatic increase among temporary- visa postdocs. For the entire cohort, temporary-visa postdocs had a first-author publication rate of 0.70 (R2 = 0.7615), which was twice that of US-native postdocs (0.34, R2 = 0.7908). Between postdoc years 2 to 5, the rate of first-author publication for temporary visa holder postdocs (0.31, R2 = 0.9552) was also greater than the rate of US- native postdocs (0.18, R2 = 0.6484).
    • Chart 7: First-author paper publication rate Chart 8: First-author publications: Caucasian vs. Asian citizen postdocs 10 9 US pub 100.0% 8 TV pub 80.0% 7 60.0% 6 % cohort Number of postdocs 40.0% papers 5 produced 20.0% 4 3 0.0% -1 -2 -3 -4 -5 -6 -7 2 Y0 Y1 Y2 Y3 Y4 Y5 Y6 Caucasian 1 Postdoc tenure (years) Asian 0 1 2 3 4 5 6 7 8 9 10 0- 1- 2- 3- 4- 5- 6- 7- 8- 9- Y Y Y Y Y Y Y Y Y Y Postdoc tenure (years) Chart 10: First-author publication rates: Asian citizen postdocs Chart 9: First-author publication Asian Asian native Asian natural Asian PR rates: published Caucasian vs. Asian citizen postdocs 5 5 per productive Publications 4 Publicatio 4 postdoc ns per 3 3 productiv 2 2 e postdoc 1 1 0 0 1 2 3 4 5 6 7 0- 1- 2- 3- 4- 5- 6- Y Y Y Y Y Y Y -1 -3 -5 -7 Y0 Y2 Y4 Y6 Postdoc tenure (years) Postdoc tenure (years)
    • US Native Demographics Caucasians and Asians All US citizens (including native, naturalized, and permanent residents) were also analyzed for any demographic differences in postdoc publication rates. Focusing on data from majority groups (specifically Caucasian and Asian populations), we sought to determine the publication rates of these populations. The survey revealed that 1250 of 2147 Caucasians (58.2%) and 217 of 361 Asians (60.1%) had self-reported first-author publications. Chart 8 shows the Table 5. First-Author Publication Rates by US-citizen Demographics percentage of Caucasian and Asian postdocs with a first-author 0-1 1-2 2-3 3-4 y 4-5 Sum y y y y 0-5y publication by postdoc tenure. African-Americans Interestingly, proportionately Unpublished 26 16 9 2 3 56 Published 7 11 10 8 3 39 fewer Asian-citizen postdocs Publications 8 16 14 15 6 59 between years 1 and 3 publish Productive 1.14 1.45 1.40 1.875 2.00 1.51 Publication Rate first-author papers compared to Hispanic/Latino Americans Caucasian citizens, but Asian Unpublished 26 15 4 6 0 51 citizens outpublish Caucasians Published 9 19 10 7 6 51 Publications 12 36 24 19 14 105 beyond postdoc year 4. The first- Productive 1.33 1.89 2.40 2.71 2.33 2.06 author publication rate among Publication Rate Native Americans Caucasian American citizens was Unpublished 3 3 0 0 0 6 0.2248 papers per postdoc year Published 1 7 4 2 2 16 2 Publications 2 13 6 5 3 29 (R = 0.9045) in this sample. Productive 2.00 1.86 1.50 2.50 1.5 1.81 Publication Rate Pacific Islander Americans We also looked at 224 Asian Unpublished 2 10 1 0 4 citizens (those not with Published 2 11 2 0 6 permanent residency status) to see Publications 3 11 5 0 10 Productive 1.00 1.50 1.00 2.50 0 1.67 if there were a difference between Publication Rate naturalized and native-born Asian Americans Unpublished 50 47 27 9 7 140 citizen productivity. Removing Published 39 31 40 42 41 193 the permanent residents resulted Publications 75 99 104 96 136 510 Productive 3.19 1.92 2.60 2.29 3.32 2.64 in a much lower publication rate Publication Rate than if they were included (Chart Caucasian Americans 10). Regression analysis did not Unpublished 264 379 130 75 33 881 Published 344 211 279 219 137 1190 produce a statistically reliable Publications 621 349 542 513 308 2333 overall publication rate for any of Productive 1.65 1.81 1.94 2.34 2.25 1.96 Publication Rate these populations, but the publication rate of US citizens (naturalized and native) in our sample was 0.0620 papers per postdoc year between years 0 and 4 (R2 = 0.82). Underrepresented Minorities Underrepresented minority postdocs had a relatively smaller publication rate compared to majority counterparts. Specifically, 18 of 24 Native American postdocs (75.0%), 6 of 10 Pacific Islander postdocs (60.0%), 63 of 115 Hispanic/Latino postdocs (54.8%), and 42
    • of 100 African-American postdocs (42.0%) claim to have authored at least one first- author publication. Table 5 shows the breakdown by postdoc year. The publication rate for published minority postdocs appears to be lower compared to published majority postdocs. Only 41% of African-American postdocs surveyed had a first-author publication compared to 50% of Latinos, 60% of Pacific Islanders, and 72.7% of Native Americans. Published Hispanic/Latino Americans have a publication rate of 0.465 papers per postdoc year (between years 0 and 4, R2 = 0.9847), while published African-American postdocs create 0.215 papers per published postdoc year (R2 = 0.9091). Gender and Children Among postdocs who have served for no more than 6 years, the overall percentage of first-author published postdocs was not significantly different between women (1318 of 3627, 36.3%) and men (1969 of 5099, 38.6%). This trend has also been observed with non-first-author publications (data not shown) from women (1144 of 3318, 34.5%) and men (1727 of 4669, 37.0%). Caring for children did not seem to adversely affect the overall publication ratio (2109 of 5805, 36.3% no children vs. 1664 of 4460, 37.3% children). Published women respondents had a generally lower ratio of publications and rate of publication of first-author articles compared to their male counterparts. Regardless of type of institution, the number of first-author papers published by women was significantly lower among published women compared to published men. The disparity becomes more dramatic when comparing male and female postdocs who claim to write three or more first-author papers: over 40% of postdocs who author one or two first- author papers are women, but the proportion is 36% women who author three first-author papers and 32% who author four or more. Between years 1 and 5, the publication rate was 0.35 papers per male published postdoctoral year (R2 = 0.9749) compared to 0.16 papers per female published postdoctoral year (R2 = 0.8936). For postdocs without children between years 1 and 5, male postdocs published 0.30 papers per year (R2 = 0.8980) while women published at 0.17 papers per year (R2 = 0.8279). For postdocs with children between years 1 and 5, male postdocs had a publication rate of 0.34 papers per year (R2 = 0.9738) while women postdocs published at 0.11 papers per year (R2 = 0.5113). Table 6 shows the persistence of this ratio compared to postdoc tenure. These data are consistent with a few studies showing that married men tend to be more “productive” in academe than women or unmarried men
    • Table 6. First-Author Publication Rates by Gender and Children Status Gender, children, first-author papers 0-1y 1-2y 2-3y 3-4y 4-5y 5-6y 0-6y Male no children 562 580 405 267 141 60 2015 Published postdocs 224 329 280 212 110 52 1207 Percentage published postdocs 28.5% 36.2% 40.9% 44.3% 43.8% 46.4% 37.5% Publications claimed 418 632 643 588 305 173 2759 Publications per published postdoc 1.87 1.92 2.30 2.77 2.77 3.33 2.29 Male with children 224 242 244 180 150 75 1115 Published postdocs 109 136 182 141 127 67 762 Percentage published postdocs 32.7% 36.0% 42.7% 43.9% 45.8% 47.2% 40.6% Publications claimed 259 300 493 411 415 368 2246 Publications per published postdoc 2.38 2.21 2.71 2.91 3.27 5.49 2.95 All male 786 822 649 447 291 135 3130 Published postdocs 333 465 462 353 237 119 1969 Percentage published postdocs 29.8% 36.1% 41.6% 44.1% 44.9% 46.9% 38.6% Publications claimed 677 932 1136 999 720 541 5005 Publications per published postdoc 2.03 2.00 2.46 2.83 3.04 4.55 2.54 Female no children 486 489 315 215 117 59 1681 Published postdocs 159 230 211 159 93 50 902 Percentage published postdocs 24.7% 32.0% 40.1% 42.5% 44.3% 45.9% 34.9% Publications claimed 246 395 441 347 208 145 1782 Publications per published postdoc 1.55 1.72 2.09 2.18 2.24 2.90 1.98 Female with children 117 126 134 118 95 38 628 Published postdocs 40 74 94 94 84 30 416 Percentage published postdocs 25.5% 37.0% 41.2% 44.3% 46.9% 44.1% 39.8% Publications claimed 82 147 197 185 200 81 892 Publications per published postdoc 2.05 1.99 2.10 1.97 2.38 2.70 2.14 All female 603 615 449 333 212 97 2309 Published postdocs 199 304 305 253 177 80 1318 Percentage published postdocs 24.8% 33.1% 40.5% 43.2% 45.5% 45.2% 36.3% Publications claimed 328 542 638 532 408 226 2674 Publications per published postdoc 1.65 1.78 2.09 2.10 2.31 2.83 2.03
    • First-Author Publications by Institution Type Does the institutional environment affect the publication rate of postdocs? To answer this question, we divided our data to see what proportion of postdocs had first-author and other peer-reviewed publications (Table 7). While 62% of our sample (3586 of 5806) claimed at least one first-author publication, government postdocs had a higher proportion (348 of 508, 69%) while private academic institution postdocs had the lowest proportion (1397 of 2361, 59%). Roughly 35% of all postdocs (2028 of 5806) claimed at least two first-author articles; public institutions in this case had proportionally more postdocs with two first-author publications (652 of 1663, 39%), and medical institutions had the smallest proportion (110 of 370, 30%). Consequently medical institutions and private academic institutions have the lowest first-author publication rates (1.34 and 1.35 publications per overall postdoc population respectively), while government and public academic institution postdocs had the highest rates (1.54 and 1.73 respectively). If one looks at just the published postdocs (those with at least one first-author publications), medical institution and government postdocs have the smallest number of publications per published postdoc (2.19 and 2.25 respectively) while research institutions and public academic institutions had the highest ratios (2.35 and 2.71 respectively). Table 7. Publication rates of journal articles by institution type. Pubs per Pubs per Group Pub n 1+ 2+ %n (1+) %n (2+) Pubs cohort postdoc 1+ postdoc Total Sole 5015 538 237 10.7% 4.7% 1163 0.23 2.16 Govt Sole 441 49 17 11.1% 3.9% 79 0.18 1.61 Med Inst Sole 295 21 7 7.1% 2.4% 34 0.12 1.62 Priv Ac Sole 2044 185 78 9.1% 3.8% 366 0.18 1.98 Pub Ac Sole 1454 218 104 15.0% 7.2% 557 0.38 2.56 Res Inst Sole 741 63 31 8.5% 4.2% 125 0.17 1.98 Total First 5806 3586 2028 61.8% 34.9% 8630 1.49 2.41 Govt First 508 348 188 68.5% 37.0% 784 1.54 2.25 Med Inst First 370 227 110 61.4% 29.7% 496 1.34 2.19 Priv Ac First 2361 1397 762 59.2% 32.3% 3181 1.35 2.28 Pub Ac First 1663 1062 652 63.9% 39.2% 2882 1.73 2.71 Res Inst First 857 524 297 61.1% 34.7% 1234 1.44 2.35 Total NonFirst 5475 3157 1821 57.7% 33.3% 7693 1.41 2.44 Govt NonFirst 485 318 198 65.6% 40.8% 875 1.80 2.75 Med Inst NonFirst 338 181 104 53.6% 30.8% 444 1.31 2.45 Priv Ac NonFirst 2237 1241 705 55.5% 31.5% 2900 1.30 2.34 Pub Ac NonFirst 1558 929 518 59.6% 33.2% 2266 1.45 2.44 Res Inst NonFirst 812 460 282 56.7% 34.7% 1138 1.40 2.47 Total Not PR 5088 2314 1510 45.5% 29.7% 7706 1.51 3.33 Govt Not PR 438 204 133 46.6% 30.4% 698 1.59 3.42 Med Inst Not PR 297 116 66 39.1% 22.2% 350 1.18 3.02 Priv Ac Not PR 2084 897 592 43.0% 28.4% 2912 1.40 3.25 Pub Ac Not PR 1481 753 518 50.8% 35.0% 2748 1.86 3.65 Res Inst Not PR 745 319 187 42.8% 25.1% 932 1.25 2.92
    • Gender differences in publication rates within institution type were also analyzed, with the data for years 0-6 of postdoctoral tenure shown in Table 8. While publication rate data are not conclusive for government, research institutes, and medical institutes, the data from academic institutions shows that women appear to have a lower rate of publishing first-author papers (0.25 papers per published public academic postdoc year [R2 = 0.9116], 0.18 papers per private academic postdoc year [R2 = 0.9269]) compared to men (0.49 papers per published public academic postdoc year [R2 = 0.9954], 0.44 papers per private academic postdoc year [R2 = 0.9696]). Table 8. Proportion of postdocs with self-reported number of first-author publications, by type of training institution and gender. Institution Gender 0 first-author 1 first-author 2 first-author 3 first-author 4+ first author Government Male 82 84 46 31 33 Female 79 83 47 18 27 (Percent of inst) (49%) (50%) (50%) (37%) (45%) Private Male 525 363 204 121 154 Academic 444 284 176 60 71 Female (46%) (44%) (46%) (33%) (32%) Public Male 323 250 176 95 156 Academic 280 171 112 61 60 Female (46%) (41%) (39%) (39%) (28%) Medical Institute Male 70 70 29 16 23 70 47 26 9 13 Female (50%) (40%) (47%) (36%) (36%) Research Male 194 127 104 50 61 Institute 143 101 52 31 27 Female (42%) (44%) (33%) (38%) (33%)
    • Table 9 shows how the publication rates vary with postdoctoral tenure and research institution type. The rates of postdocs in their first five years of training are remarkably similar. It appears that government and private academic postdocs had a smaller first- author publication rate between postdoctoral years 2 to 5 (government 0.256 first-author publications per postdoc year, R2 = 0.6476; private academic 0.298 publications per postdoc year, R2 = 0.9557). In contrast, medical institutes and public academic institutions had the highest productivity rates (medical institute 0.500 first author publications per postdoc year, R2 = 0.9314; 0.445 publications per postdoc year, R2 = 0.9791). Table 9. Publication rates of male postdocs with first-author peer-reviewed papers, by training institution and postdoctoral tenure (within the first six years). Publications per Published First-author published Institution Male Postdocs postdocs Percentage publications postdoc Public Academic Y0-1 239 99 41.4% 198 2.00 Public Academic Y1-2 266 163 61.3% 333 2.04 Public Academic Y2-3 197 150 76.1% 366 2.44 Public Academic Y3-4 125 108 86.4% 320 2.96 Public Academic Y4-5 70 62 88.6% 217 3.50 Public Academic Y5-6 41 38 92.7% 272 7.16 Public Academic Y0-6 total 938 620 66.1% 1706 2.75 Private Academic Y0-1 366 151 41.3% 238 1.58 Private Academic Y1-2 320 185 57.8% 341 1.84 Private Academic Y2-3 262 177 67.6% 420 2.37 Private Academic Y3-4 172 132 76.7% 342 2.59 Private Academic Y4-5 124 99 79.8% 320 3.23 Private Academic Y5-6 46 39 84.8% 188 4.82 Private Academic Y0-6 total 1290 783 60.7% 1849 2.36 Government Y0-1 43 23 53.5% 38 1.65 Government Y1-2 64 33 51.6% 70 2.12 Government Y2-3 50 38 76.0% 61 1.61 Government Y3-4 46 33 71.7% 81 2.45 Government Y4-5 34 29 85.3% 64 2.21 Government Y5-6 4 4 100.0% 12 3.00 Government Y0-6 total 241 160 66.4% 326 2.04 Medical Institute Y0-1 42 22 52.4% 50 2.27 Medical Institute Y1-2 47 24 51.1% 34 1.42 Medical Institute Y2-3 40 29 72.5% 52 1.79 Medical Institute Y3-4 31 24 77.4% 64 2.67 Medical Institute Y4-5 24 18 75.0% 36 2.00 Medical Institute Y5-6 10 7 70.0% 33 4.71 Medical Institute Y0-6 total 194 124 63.9% 269 2.17 Research Institute Y0-1 99 39 39.4% 81 2.08 Research Institute Y1-2 131 66 50.4% 133 2.02 Research Institute Y2-3 103 68 66.0% 194 2.85 Research Institute Y3-4 83 65 78.3% 167 2.57 Research Institute Y4-5 42 33 78.6% 79 2.39 Research Institute Y5-6 32 30 93.8% 107 3.57 Research Institute Y0-6 total 490 301 61.4% 761 2.53
    • Table 10. Publication rates of female postdocs with first-author peer-reviewed papers, by training institution and postdoctoral tenure (within the first six years). Publications per Published First-author published Institution Female Postdocs postdocs Percentage publications postdoc Public Academic Y0-1 163 57 35.0% 96 1.68 Public Academic Y1-2 192 98 51.0% 186 1.90 Public Academic Y2-3 149 105 70.5% 241 2.30 Public Academic Y3-4 77 61 79.2% 141 2.31 Public Academic Y4-5 46 35 76.1% 95 2.71 Public Academic Y5-6 23 22 95.7% 70 3.18 Public Academic Y0-6 total 650 378 58.2% 829 2.19 Private Academic Y0-1 278 87 31.3% 142 1.63 Private Academic Y1-2 253 129 51.0% 222 1.72 Private Academic Y2-3 166 116 69.9% 218 1.88 Private Academic Y3-4 147 102 69.4% 201 1.97 Private Academic Y4-5 94 76 80.9% 176 2.32 Private Academic Y5-6 38 30 78.9% 72 2.40 Private Academic Y0-6 total 976 540 55.3% 1031 1.91 Government Y0-1 36 16 44.4% 29 1.81 Government Y1-2 52 29 55.8% 47 1.62 Government Y2-3 56 40 71.4% 79 1.98 Government Y3-4 48 23 47.9% 82 3.57 Government Y4-5 28 25 89.3% 57 2.28 Government Y5-6 10 7 70.0% 23 3.29 Government Y0-6 total 230 140 60.9% 317 2.26 Medical Institute Y0-1 37 7 18.9% 9 1.29 Medical Institute Y1-2 35 16 45.7% 21 1.31 Medical Institute Y2-3 32 19 59.4% 39 2.05 Medical Institute Y3-4 21 16 76.2% 30 1.88 Medical Institute Y4-5 16 5 31.3% 16 3.20 Medical Institute Y5-6 9 8 88.9% 24 3.00 Medical Institute Y0-6 total 150 71 47.3% 139 1.96 Research Institute Y0-1 92 35 38.0% 57 1.63 Research Institute Y1-2 87 35 40.2% 67 1.91 Research Institute Y2-3 54 33 61.1% 71 2.15 Research Institute Y3-4 43 39 90.7% 80 2.05 Research Institute Y4-5 31 28 90.3% 48 1.71 Research Institute Y5-6 10 7 70.0% 21 3.00 Research Institute Y0-6 total 317 177 55.8% 344 1.94
    • Disciplinary Differences in Publication Rate Responses were also separated by self-reported discipline to see whether publication rates for physical science postdocs differed from postdocs in biomedical science who constitute the majority of responses in the survey. The results for many of these areas are shown in Table 10. Calculations of publication rate show that physics and chemistry postdocs produce around 1 paper per postdoc year for all cohort postdocs, just over 1 paper for each postdoc with a publication. Math postdocs have a publication rate of around 0.7 papers per year, while earth science and computer science postdocs have a rate of around 0.8 to 1.0 papers per year. This rate is much higher than the rate of publication for molecular biology postdocs (0.4 to 0.5 papers per postdoc year). Many scientific workforce studies are concerned about the lack of women who make it to faculty positions in the physical (non-biomedical) sciences. Table 11 looks at the respondents from the Sigma Xi survey that reported their first-author publications in the areas of chemistry, physics, and ecology/evolutionary biology. The ratio of first-author publications to published postdocs remains higher for men compared to women between years 1-4 of training. Remarkably, the number of women respondents in physics and chemistry drop off is dramatically smaller after two years of postdoctoral training, suggesting that early postdoctoral mentoring for the purposes of research productivity may be a critical area of intervention in order to retain women in the faculty pool. Table 11. Publication rate of non-biomedical science postdocs, by gender and year of training (up to 6 years) Discipline Gender Year Postdocs Published Percentage First-author Pubs per Postdocs Published publications published Postdocs claimed postdoc Chemistry Male 0-1 96 38 39.6% 88 2.32 1-2 96 56 58.3% 101 1.80 2-3 66 48 72.7% 118 2.46 3-4 30 24 80.0% 69 2.88 4-5 21 16 76.2% 76 4.75 5-6 9 9 100.0% 57 6.33 Total 0-6 318 191 60.1% 509 2.66 Chemistry Female 0-1 37 14 37.8% 22 1.57 1-2 36 20 55.6% 34 1.70 2-3 17 15 88.2% 30 2.00 3-4 7 5 71.4% 8 1.60 4-5 6 6 100.0% 32 5.33 5-6 1 1 100.0% 1 1.00 Total 0-6 104 61 58.7% 127 2.08 Physics Male 0-1 64 29 45.3% 39 1.34 1-2 56 36 64.3% 83 2.31 2-3 45 36 80.0% 120 3.33 3-4 22 19 86.4% 80 4.21 4-5 13 10 76.9% 63 6.30 5-6 4 4 100.0% 29 7.25 Total 0-6 204 134 65.7% 414 3.09 Physics Female 0-1 11 6 54.5% 10 1.67 1-2 21 14 66.7% 49 3.50 2-3 4 2 50.0% 7 3.50 3-4 2 2 100.0% 5 2.50 4-6 3 3 100.0% 18 6.00
    • Total 0-6 41 27 65.9% 89 3.30 Ecology and Male 0-1 37 27 73.0% 59 2.19 Evolutionary Biology 1-2 26 18 69.2% 37 2.06 2-3 12 11 91.7% 39 3.55 3-4 17 14 82.4% 51 3.64 4-6 6 6 100.0% 23 3.83 Total 0-6 61 49 80.3% 150 3.06 Ecology and Female 0-1 18 12 66.7% 20 1.67 Evolutionary Biology 1-2 22 15 68.2% 34 2.27 2-3 12 12 100.0% 29 2.42 3-4 6 5 83.3% 18 3.60 4-6 4 2 50.0% 5 2.50 Total 0-6 44 34 77.3% 86 2.53 Does mentoring affect published postdoc productivity? Respondents were also asked about the mentoring and supervision style they had with their postdoctoral training. We have previously reported that the existence of a formal mentoring plan was a key component to more effective postdoctoral training. Table 12 shows the responses and their self-reported publications. The proportion of postdocs who had a first author publication was not significantly different depending on whether the postdoc had a training plan. However, postdocs with a formal written plan tended to have more first-author publications per published postdoc. Postdocs who perceived that their written plan was useful had a higher publication ratio. Table 12. The effect of first-author publication ratio on formalized mentoring structure First-author Training Characteristic Total Published Publication Ratio Publications Formal evaluations? …Yes 1219 804 2022 2.51 …No 3714 2269 5436 2.40 Written plan 669 421 1109 2.63 … was not useful 40 24 51 2.13 … was somewhat useful 292 183 460 2.51 … was very useful 314 204 545 2.67 Oral unwritten plan 3520 2179 5192 2.38 No plan 1646 1043 2554 2.45 Frequency of advisor- trainee meetings …Daily 391 279 798 2.86 …Weekly 1660 1056 2436 2.31 …Monthly 1045 629 1408 2.24 …Yearly 284 164 365 2.23 Were advisor expectations included? …Formal written, yes 363 236 615 2.61 …Formal written, no 190 111 254 2.29 ...Oral unwritten, yes 1452 924 2231 2.41 …Oral unwritten, no 1233 727 1628 2.24
    • Discussion This is the first detailed analysis of data looking at the publication rates of postdoctoral scholars engaged in training as reported in the 2004 Sigma Xi Postdoctoral Survey. Difficulties in data interpretation are apparent when looking at the disciplinary representation of the respondents (majority biomedical sciences) and the lack of response from underrepresented minorities. Regardless, some insights may be gained from the responses we did gather on their productivity, as measured by their self-reported publication of first-author peer-reviewed papers, which are considered to be a critical measuring stick for preparation for a future career in academic sciences and engineering. In addition, the production of first-author publications can be an indirect measure to the innovative capacity of this highly educated population of scientists that will drive future scientific discovery. In general, the American postdoctoral training system produces approximately one first- author and one collaborative non-first-author publication (data not shown) every two years. Around 40% of postdocs report they have a first-author publication within their first year of training, but it is not clear whether the publication is related to their graduate thesis work or their “new” work as a postdoc. However, as postdocs continue to be trained, it appears that it takes roughly 12 to 18 months to produce a first-author publication and slightly longer for a second first-author publication. This timing suggests that for most postdocs, postdoctoral training does not necessarily result in more efficient production of innovative discoveries as evidenced by the publication of first-author papers. It is possible the delay could be due to collecting more data for a higher-impact first-author publication that would be more valued in an academic job search or investigating different angles until collecting enough data for a quality publication. Temporary visa postdocs report a higher publication rate of first-author peer-reviewed papers compared to US native-born postdocs (roughly twice as productive). By the fourth postdoctoral training year, there were twice as many temporary-visa postdocs with first-author publications compared to the number of US native postdocs with first-author papers. Though very small in number, more postdocs on temporary visa are engaged in extended postdoctoral training (beyond 6 years). Among US native postdocs, the calculated publication rate of majority populations (Caucasian and Asians) was generally higher than the publication rates from underrepresented minorities. While the number of respondents make it difficult to form conclusions, we do observe that Hispanic-heritage postdocs apparently produce 0.47 first- author papers a year, which is a higher rate than Caucasians. In contrast, African- American postdocs produce 0.22 first-author papers a year. It is not known why this disparity in publication rate exists. There is also a gender gap in first-author publication productivity. While the proportion of women with first-author publications is not much different than men in the postdoctoral cohort, the rate of publication per training year is half the rate that men produce first-author publications. Caring for children did not appear to affect the
    • proportion of postdocs with first-author papers, but there may be a subtle difference in publication rate, particularly among postdoctoral women. There seemed to be little major difference in production of first-author papers by type of research institution. Government institutions had the highest proportion of postdoctoral scientists with 1 or more first-author and secondary-author publications. Disciplinary differences in publication rate reveal that our overall measurement of first author publication productivity reflects the majority biomedical science population. Indeed, postdocs in areas outside the biomedical sciences seem to produce more quickly first-authored papers. This is likely due to the differences in the nature of research in these areas. Many of these disciplines do not have a tradition of postdoctoral training unless the scholars intend to pursue an academic professorial position. It is not clear whether this higher rate of publication results in more satisfied postdocs who attain their professional goals in academia, industry, or elsewhere, but it can be surmised that the nature of their work and disciplinary environment may be supportive for these postdocs. That said, the percentage of women in non-biomedical scientific postdoctoral positions is still extremely low. Not only are the women in this cohort reporting relatively lower rates of publication but their numbers decrease dramatically after the second year of postdoctoral training. If the numbers in this report are representative of those disciplines, it is not surprising that few women candidates exist that could meet the disciplinary threshold of “quality” if it were measured solely by first-author publication productivity. The limitation of this study is that individual respondents were not asked to disclose the value of their first-author publications. While the number of first-author papers is important, in any job interview, the quality, impact, and citation values of those and other secondary-author publications is also deemed to be critical in determining the caliber of a scientist-candidate for fellowships or job positions. Implications for Training the Scientific Workforce Much attention has been brought recently to the dearth of quality candidates for principal investigator or scientist positions for academic or non-academic positions. Although much attention has been brought upon the climate that dissuades women and minorities from pursuing research careers, aside from counting individuals, little is really known about the output of the American postdoctoral training system. Indeed, one of the most frustrating statements in this debate is that we do not have enough scientists who are ready for these types of jobs. Standards that had guided many doctoral students are no longer the case with postdoctoral training. While there are specific guidelines and milestones set by curricular and thesis committees (such as the publication of at least one first-author paper), postdoctoral training relies much more on fulfilling personal and professional expectations of the trainee and the mentor. A lack of oversight or quality control puts the postdoctoral researcher at greater professional vulnerability, and the lack of stated professional landmarks to make a trainee a successful scientist often can result in a less- than-optimal professional experience.
    • While first-author publications cannot by itself be the sufficient measure of a successful postdoctoral fellowship, it is a critical metric for the review of any scientist’s productivity or potential in a grant review or job application. While this study is not able to take impact of publications into account (such as number of citations to each paper), the number of publications including first-author publications is reliably a quantitative metric that narrows an applicant pool. Where that minimum number is set may vary greatly among disciplines, institutions, and departments, but the goal for most postdoctoral trainees should still be to place themselves in a competitive position. It is conceded that the institutions represented in this survey are responsible for training a majority of postdoctoral scholars whose intentions are to pursue careers in academia at research intensive and undergraduate teaching institutions. However, these same institutions will most likely hire those postdoctoral scholars, and are somewhat responsible for the quality of the hiring pool for themselves as well as industry positions, just as they would be responsible for the pools of undergraduate candidates who will apply for graduate programs in medicine, law, or “the workforce.” If the minimum number of papers for a “quality” candidate is 5 first-author papers, these data suggest that our postdoctoral training system is woefully inadequate to seed the pool with diverse quality candidates that may be able to reach that goal. According to these data, higher numbers favor postdocs who are not American citizens and who are Caucasian male postdocs. For disciplines that tend to have higher publication rates, such as physics and chemistry, the bar seems to be set at a point where women could likely be summarily excluded because their publication rate could not keep up with their male counterparts. As a result the women who remain in postdocs tend to be as well-published as the men, but their proportion in the pool drastically decreases. Those women who publish fewer first-author publications may have to create much higher-impact first-author papers compared to men to maintain a viable candidacy. Thus, women could also be more likely to opt-out of a career track towards investigatorship as personal work-life issues could distract them from maintaining a competitive publication rate. The publication rate may also explain the expansion of postdoctoral tenure for biomedical scientists. The data suggest it takes almost a year to produce a first-author paper, so to reach the critical “five paper” threshold, the average postdoc would have to serve five or more years in biomedical sciences to be a viable candidate. Since the average age of a postdoc (in their second year of training) is around 33, the age for a scientist who completed postdoctoral training would be in the mid-to-late 30’s. However, competitive applicants are clearly not the “average” candidates; hiring decisions favor extraordinary scientists who can produce high-impact research and get funded quickly. These data suggest that there is a great disconnect between the development of postdoctoral scholars at the institutional and disciplinary level and the preparation for viable faculty candidates “of quality”, especially in non-biomedical scientific areas. It is thus important that institutions that serve postdoctoral scholars be prepared to collect critical data to properly assess and improve the postdoctoral training system. It should be in the best interests of academic institutions to have their postdocs become more
    • informed about the competitive nature of a career as a principal investigator and make themselves more accountable to producing more productive candidates. Our data show that modest changes in the structure of a postdoctoral supervision could correspond to higher publication rates. Simple steps such as requiring and reviewing formal written plans, explicitly stating the roles for the scholar and the supervisor, and scheduling more interactions and meetings could result in a greater positive outcome in postdoctoral publication rate. These interventions also may be critical to supporting the career paths of women and underrepresented minority scientists at the postdoctoral stage. It is also clear that an outcomes-oriented postdoctoral training system may be desirable but does not address professional climate issues. There is increased concern about the production of postdoctoral scientists who would not act responsibly in their conduct of research in order to get that extra first-author credit. Having more postdoctoral scientists who publish papers may be desirable if we were to measure scientific innovation, but first-author papers do not address the lack of job preparation and the difficult transition for these same postdoctoral scientists to non-PI-track positions. Finally, lifestyle choices play a role in the decision to stay on the research path independent of the number of publications an individual has produced. In summary, this paper provides insight into the state of postdoctoral training in the United States and a measurement of the output it generates in the form of first-author publications. As the research infrastructure and funding for graduate and postgraduate scientists improves, it is hoped that the publication rates of the postdoctoral population can improve. However, interventions that will retain women and minorities in the scientific workforce pipeline must keep the publication data in mind when developing innovative policies and strategies to close the representation gap. Our data shed some insight into the effects of more structured and formalized mentoring as a possible target for future assessment of interventions for postdocs. Acknowledgments This paper would not have been possible without the leadership of the Alfred Sloan Foundation and Sigma Xi to underwrite the 2004 Postdoctoral Survey, and I hope that there will be a commitment to produce more frequent surveys and analysis teams to chart progress in this and other areas. Specific credit goes to the survey project director Geoff Davis and his program associate Jenny Zilaro, with whom I worked closely to administer the survey at Duke University with Dean Jo Rae Wright, Associate Provost Jim Siedow, and Provost Peter Lange. I thank Alyson Reed, Amber Budden, and my Diversity Committee colleagues from the National Postdoctoral Association for allowing me to do further analysis with this data. I also recognize the support of my colleagues at George Mason University and Case Western Reserve University for preliminary feedback on these data. This paper is written in memory of and honor to Deborah Swope (NIEHS) who was among the earliest advocates in support for the survey.