Funding Disparities in Science Careers

Nonparametric t tests and Kruskal-Wallis tests with posthoc Dunn correction for multiple comparisons were used to determine significance between median funding dollar amounts across groups. Multivariable logistic regression was used to determine the relative odds of women and investigators from underrepresented racial and ethnic groups of being an SPI compared with men and White investigators, respectively. Covariates include PI’s highest degree and career stage. Degree was defined as MD, MD/PhD, PhD, or other degrees. PI’s career stage was approximated using investigator’s age, categorized as early (age under 46 years), middle (age 46 to 58 years), and late (age above 58 years), as described previously.^{1 (link)} Finally, we included 3 time periods that delineate significant changes in the NIH budget: 1991-1998 (phase 1) before the first budget increase, 1999-2014 (phase 2) between the first and second budget increase, and 2015-2020 (phase 3) after the second budget increase, and tested the interaction between phase and gender, ethnic, and racial identities to determine whether the relative odds of being an SPI for disadvantaged groups (eg, women, Black, and Hispanic) have changed over time. Adjusted percentage of SPI investigators within each combined subgroup of gender, ethnic, and racial identity was determined from the fully adjusted logistic models. Statistical tests were 2-sided with type 1 error rate of 0.05. All analyses were performed using Stata version 16.1 (Stata Inc).