4 Comments
Last year, we showed that there were slight improvements in the funding and success rates for Black/African American and Hispanic researchers in fiscal year (FY) 2022. The race-ethnicity gaps also narrowed over the past few years, though the overall number of Black/African American and Hispanic applicants remained low. We are updating these analyses for FY23 and present some findings below (full report available here).
Similar to our prior analyses, funding rates for FY23 focused on new (Type 1) Research Project Grant (RPG) and R01-equivalent applicants according to the race-ethnicity of designated Principal Investigators (PIs). Of note:
- Race and ethnicity data were obtained from self-reported information PIs provided on their eRA profiles. This Nexus article explains more about these (and other) demographic data and how they are collected and used at NIH.
- NIH discusses application-based metrics (award rates and success rates) as well as person-based metrics (funding rates) in specific situations (such as through our blog or NIH Data Book).
- PIs are referred to as applicants or awardees for simplicity here. Please understand though that applicants and awardees are the organizations that designate PIs.
- Native Hawaiian or Other Pacific Islanders as well as American Indian or Alaska Native researchers were excluded from these analyses due to small cell sizes. We do not publicly report sample sizes that are sufficiently small (<12), which is considered as potentially identifiable.
Table 1 shows the race and ethnicity of PIs on at least one new RPG application in FY23. Compared with White and Asian scientists, Black/African American scientists were more likely to be women, hold an M.D. degree, propose a research idea involving human participants, and be an early-stage investigator.
Table 1: Characteristics by race-ethnicity of scientists who were designated as a Principal Investigator on at least one Type 1 FY23 Research Project Grant (RPG) application
| Characteristic | White | Asian | Unknown | Hispanic | Black | |
| Total N (%) | 22376 (54.7) | 10798 (26.4) | 4191 (10.2) | 2263 (5.5) | 1283 (3.1) | |
| Gender | Male | 13076 (58.4) | 7093 (65.7) | 1409 (33.6) | 1245 (55.0) | 661 (51.5) |
| Female | 9040 (40.4) | 3518 (32.6) | 646 (15.4) | 981 (43.3) | 597 (46.5) | |
| Unknown | 260 (1.2) | 187 (1.7) | 2136 (51.0) | 37 (1.6) | 25 (1.9) | |
| Degree | PhD | 16232 (72.5) | 7794 (72.2) | 1783 (42.5) | 1596 (70.5) | 815 (63.5) |
| MD | 3250 (14.5) | 1239 (11.5) | 400 (9.5) | 293 (12.9) | 210 (16.4) | |
| MD-PhD | 1952 (8.7) | 1310 (12.1) | 237 (5.7) | 226 (10.0) | 132 (10.3) | |
| Other | 942 (4.2) | 455 (4.2) | 1771 (42.3) | 148 (6.5) | 126 (9.8) | |
| Submitted an R01-Equivalent Application | 16782 (75.0) | 8170 (75.7) | 2700 (64.4) | 1653 (73.0) | 899 (70.1) | |
| Submitted an R21 or R03 Application | 6696 (29.9) | 3743 (34.7) | 1441 (34.4) | 732 (32.3) | 377 (29.4) | |
| Submitted an Animal Application | 10334 (46.2) | 6130 (56.8) | 1613 (38.5) | 1041 (46.0) | 374 (29.2) | |
| Submitted a Human Application | 11398 (50.9) | 4419 (40.9) | 2221 (53.0) | 1177 (52.0) | 871 (67.9) | |
| Early-Stage Investigator | 4005 (17.9) | 2058 (19.1) | 539 (12.9) | 516 (22.8) | 408 (31.8) | |
| Number Prior Type 1 RPG Applications* | Mean (SD) | 9.2 (9.5) | 10.6 (11.7) | 5.6 (9.1) | 7.7 (9.0) | 6.2 (8.5) |
| Number Prior Type 1 RPG Awards* | Mean (SD) | 1.8 (2.2) | 1.7 (2.2) | 0.9 (1.8) | 1.4 (2.0) | 0.9 (1.5) |
| Number of FY23 Applications | Mean (SD) | 1.6 (1.0) | 1.8 (1.4) | 1.5 (1.0) | 1.6 (1.1) | 1.4 (0.9) |
| Submitted from Institution of Higher Education | 18724 (83.7) | 9175 (85.0) | 3318 (79.2) | 1940 (85.7) | 1130 (88.1) | |
| Submitted from Research Organization | 1429 (6.4) | 633 (5.9) | 300 (7.2) | 134 (5.9) | 56 (4.4) | |
| Submitted from Independent Hospital | 2068 (9.2) | 912 (8.4) | 375 (8.9) | 150 (6.6) | 68 (5.3) | |
| Funded as Principal Investigator | 6882 (30.8) | 3093 (28.6) | 896 (21.4) | 643 (28.4) | 296 (23.1) |
*The numbers of prior Type 1 RPG applications and awards refers to the period of FYs 2010 to FY22.
Funding rates are next shown for PIs who sought funding on NIH RPG applications (Figure 1). The funding rate is the ratio of the number of unique applicant PIs who are successful on at least one application to the number of unique applicant PIs who submitted at least one application. The number of RPG applicants generally increased until FY21, while the number of awardees continually trended upward. The funding rate generally decreased until FY13 (the year of sequestration), at which point it began increasing each year. The other two vertical lines represent the beginning (1998) and end (2003) of the NIH budget doubling.
Figure 1: NIH RPG funding rates by fiscal year
The R01-equivalent funding rates are shown in Figure 2. The number of R01-equivalent applicants generally increases until FY21, while the number of awardees continually trended upward after FY13. The funding rate generally increases after FY13.
Figure 2: NIH R01-equivalent funding rates by fiscal year
Figure 3 shows the number of Type 1 RPG and R01-equivalent applicants by race or ethnicity. In FY23, the number of White and Asian applicants declined while the number of Black/African American and Hispanic applicants continued to increase, though their overall numbers remain low.
Figure 3: Number of PI applicants by race-ethnicity for Type 1 RPG (Panel A) and Type 1 R01-equivalent (Panel B) awards by fiscal year.
Figure 4 again shows funding rates. The gaps between White, Asian and Hispanic investigators have largely disappeared, while the gap for Black/African American investigators persist (slightly growing in FY23 compared to FY22).
Figure 4: Funding rates for Type 1 RPG (Panel A) and Type 1 R01-equivalent (Panel B) applicants by fiscal year according to race and ethnicity
Table 2 shows the results of logistic regression analyses for funding of at least one FY23 Type 1 RPG application. Black scientists were less likely to be funded, with the effect size decreasing by approximately 32% after adjusting for confounding variables. In the multivariable model, the strongest predictors of funding success were the number of applications submitted in FY23 and early-stage investigator status. The strongest predictors of funding failure were sex other than male and female, and a degree other than Ph.D., M.D.-Ph.D., or M.D.
Table 2: Logistic regression models for funding of at least one FY23 Type 1 RPG application among scientists designated as Principal Investigators.
The numbers of prior Type 1 RPG applications and awards refers to the period of FY10 to FY22. Model 1 focuses on race only. Model 2 focuses on gender only. Model 3 includes both race and ethnicity and gender variables. Values shown are regression coefficients (standard errors). A value greater than zero implies a higher likelihood of funding, while a value less than zero implies a lower likelihood.
| Dependent Variable: Funded as Principal Investigator | |||
| Model 1 | Model 2 | Model 3 | |
| Asian (vs White) | −0.101***(0.026) | −0.153*** (0.027) | |
| Race Unknown (vs White) | −0.491*** (0.040) | −0.116* (0.050) | |
| Hispanic (vs White) | −0.112* (0.049) | −0.070 (0.051) | |
| Black (vs White) | −0.393*** (0.068) | −0.267*** (0.070) | |
| Female (vs Male) | 0.003 (0.023) | 0.095*** (0.025) | |
| Gender Unknown (vs Male) | −0.804*** (0.055) | −0.284*** (0.071) | |
| MD (vs PhD) | 0.139*** (0.035) | ||
| MD-PhD (vs PhD) | 0.031 (0.039) | ||
| Other degree (vs PhD) | −0.315*** (0.058) | ||
| Submitted R01 Equivalent | 0.083* (0.035) | ||
| Submitted R21 or R03 | −0.051 (0.031) | ||
| Submitted Animal Application | −0.092*** (0.027) | ||
| Submitted Human Application | −0.103*** (0.027) | ||
| Early-Stage Investigator | 0.263*** (0.032) | ||
| Number Prior Type 1 RPG Applications | −0.034*** (0.002) | ||
| Number Prior Type 1 RPG Awards | 0.213*** (0.008) | ||
| Number FY23 Applications | 0.436*** (0.013) | ||
| Submitted from Institution of Higher Education | 0.242*** (0.051) | ||
| Submitted from Research Organization | 0.238*** (0.060) | ||
| Submitted from Hospital | 0.252*** (0.057) | ||
| Constant | −0.812*** (0.014) | −0.860*** (0.014) | −1.905*** (0.062) |
| Observations | 40,911 | 40,911 | 40,911 |
| Log Likelihood | -24,494.78 | -24,457.15 | -23,031.93 |
| Akaike Inf. Crit. | 48,999.57 | 48,920.29 | 46,105.86 |
Note: *p<0.1; **p<0.05; ***p<0.01
These analyses echo findings presented last year. There were slight improvements in funding and success rates in FY23, likely because of fewer applicants and applications. The race and ethnicity gaps continue to narrow, though the White-Black gap still persists, and increased in FY23 compared to FY22. FY23 also saw more Black and Hispanic applicants, though the total numbers remain quite small.
We are grateful to our colleagues in the NIH Office of Extramural Research Division of Statistical Analysis and Reporting (DSAR) for their help with these analyses.
I still don’t understand why you refuse to include the original bias indicator from Ginther and Hoppe in these updates. It is critical to evaluate per-application success in addition to per-applicant success to understand whether the NIH has improved, stood pat or actually worsened the substantial bias against the applications of Black PIs that was first published in 2011 and replicated almost perfectly in 2019.
I also will point out that when you talk about closing gaps, you conveniently avoid pointing out the raw numbers. One percent of Black applicants is 13 funded PIs. One percent of white applicants is 224 funded PIs. These graphs present what appear to be a one or two percent increase for white applicants, i.e., 224-448 funded PIs, over FY2022. At the same time, a one or two percent decrease for Black applicants means about 13-26 fewer funded PIs. Let’s look at this another way. The graphs you present look like about 8 percent more Black PIs being funded would equalize the rate, problematic as that is. That is only 104 more funded Black PIs which is less than half of the *increase* over FY2022 that you have funded for the white PI population.
Great points Mike. Thanks for pointing this out.
Great points, Mike. It is important to highlight this difference.
Why are New Investigators not in this model?