G 3). For comparison, the correlation between the survey-based measure of FABs and African American PhD representation over these same 18 fields was -.73 [-.89, -.41], p < .001, jasp.12117 which was not significantly different from the correlation with the brilliance buy U0126-EtOH language score, z = -1.29, p = .200. As before, however, partialling out each measure of FABs from the correlation of the other with African American PhD representation suggested that the survey-based FAB measure explains unique variance in race gaps (partial correlation for the survey-based measure adjusting for the brilliance language score: r[15] = -.61 [-.84, -.18], p = .010; partial correlationTable 1. Multiple regression analysis predicting female representation at the PhD level. Predictor STEM indicator variable Brilliance language score Hours worked (on-campus)a Systematizing vs. empathizing Selectivity Quantitative GRE R2 -.39 -.48* .26 .01 .10 -.53 t -1.27 -2.60 0.98 0.04 0.54 -1.62 77.9 p .230 .025 .348 .971 .597 .* p < .05. N = 18 disciplines. "STEM" stands for "(Natural) Science, Technology, Engineering, and Mathematics." Although Leslie, Cimpian, et al. [1] collected data on the number of hours worked off campus as well, they found that the number of hours worked on campus was a better predictor of female representation than the total number of hours worked. Thus, to be conservative, we included this stronger competitor in our regression analyses. However, the brilliance language score remains a significant predictor even when the total number of hours worked (on- plus off-campus) is used in the regression. doi:10.1371/journal.pone.0150194.taPLOS ONE | DOI:10.1371/journal.pone.0150194 March 3,9 /"Brilliant" "Genius" on RateMyProfessors Predict a Field's DiversityFig 3. Use of the words "brilliant" SART.S23506 and “genius” on RateMyProfessors.com predicts the proportion of 2011 U.S. PhDs who are African American. doi:10.1371/journal.pone.0150194.gfor the brilliance language score adjusting for the survey-based measure: r[15] = -.14 [-.58, .37], p = .596). Notably, the brilliance language score remained a significant predictor of race gaps in PhD representation when adjusting for a field’s work demands, selectivity, and average Quantitative GRE scores, = -.65 [-1.15, -0.14], p = .016 (see Table 2). None of these controls were themselves significant in the model. Regression models using the separate brilliance language scores computed from male and female Nutlin-3a chiral solubility instructors’ evaluations found these scores to also explain unique variance in African Americans’ PhD representation (see Table E in S1 File). It is worth noting that the relationship between brilliance-related language on RateMyProfessors.com and African Americans’ PhD representation speaks against a possible alternative interpretation of the results concerning women’s representation: Perhaps fields that have more mentions of “brilliant” and “genius” in their online evaluations do so just because more undergraduate men take courses in them, and men may be more likely than women to value and comment on these traits (whereas women may be correspondingly more focused on the levelTable 2. Multiple regression analysis predicting African American representation at the PhD level. Predictor STEM indicator variable Brilliance language score Hours worked (on-campus) Selectivity Quantitative GRE R2 -.32 -.65* -.20 -.37 -.09 t ?.79 ?.80 ?.53 -1.40 ?.25 49.0 p .447 .016 .607 .186 .* p < .05. N = 18 disciplines. The brilliance language scor.G 3). For comparison, the correlation between the survey-based measure of FABs and African American PhD representation over these same 18 fields was -.73 [-.89, -.41], p < .001, jasp.12117 which was not significantly different from the correlation with the brilliance language score, z = -1.29, p = .200. As before, however, partialling out each measure of FABs from the correlation of the other with African American PhD representation suggested that the survey-based FAB measure explains unique variance in race gaps (partial correlation for the survey-based measure adjusting for the brilliance language score: r[15] = -.61 [-.84, -.18], p = .010; partial correlationTable 1. Multiple regression analysis predicting female representation at the PhD level. Predictor STEM indicator variable Brilliance language score Hours worked (on-campus)a Systematizing vs. empathizing Selectivity Quantitative GRE R2 -.39 -.48* .26 .01 .10 -.53 t -1.27 -2.60 0.98 0.04 0.54 -1.62 77.9 p .230 .025 .348 .971 .597 .* p < .05. N = 18 disciplines. "STEM" stands for "(Natural) Science, Technology, Engineering, and Mathematics." Although Leslie, Cimpian, et al. [1] collected data on the number of hours worked off campus as well, they found that the number of hours worked on campus was a better predictor of female representation than the total number of hours worked. Thus, to be conservative, we included this stronger competitor in our regression analyses. However, the brilliance language score remains a significant predictor even when the total number of hours worked (on- plus off-campus) is used in the regression. doi:10.1371/journal.pone.0150194.taPLOS ONE | DOI:10.1371/journal.pone.0150194 March 3,9 /"Brilliant" "Genius" on RateMyProfessors Predict a Field's DiversityFig 3. Use of the words "brilliant" SART.S23506 and “genius” on RateMyProfessors.com predicts the proportion of 2011 U.S. PhDs who are African American. doi:10.1371/journal.pone.0150194.gfor the brilliance language score adjusting for the survey-based measure: r[15] = -.14 [-.58, .37], p = .596). Notably, the brilliance language score remained a significant predictor of race gaps in PhD representation when adjusting for a field’s work demands, selectivity, and average Quantitative GRE scores, = -.65 [-1.15, -0.14], p = .016 (see Table 2). None of these controls were themselves significant in the model. Regression models using the separate brilliance language scores computed from male and female instructors’ evaluations found these scores to also explain unique variance in African Americans’ PhD representation (see Table E in S1 File). It is worth noting that the relationship between brilliance-related language on RateMyProfessors.com and African Americans’ PhD representation speaks against a possible alternative interpretation of the results concerning women’s representation: Perhaps fields that have more mentions of “brilliant” and “genius” in their online evaluations do so just because more undergraduate men take courses in them, and men may be more likely than women to value and comment on these traits (whereas women may be correspondingly more focused on the levelTable 2. Multiple regression analysis predicting African American representation at the PhD level. Predictor STEM indicator variable Brilliance language score Hours worked (on-campus) Selectivity Quantitative GRE R2 -.32 -.65* -.20 -.37 -.09 t ?.79 ?.80 ?.53 -1.40 ?.25 49.0 p .447 .016 .607 .186 .* p < .05. N = 18 disciplines. The brilliance language scor.
Posted inUncategorized