Table 5 Coefficient estimates—second expert panel evaluation.

From: Gender diversity of research consortia contributes to funding decisions in a multi-stage grant peer-review process

 

(1)

(2)

(3)

Female ratio

0.097

0.010

−0.089

 

(0.197)

(0.181)

(0.191)

Gender balanced

0.019

0.033

0.077

 

(0.072)

(0.075)

(0.093)

Reviewers scores mean

0.685***

0.693***

0.708***

 

(0.109)

(0.114)

(0.106)

Reviewers scores variance

0.008

−0.000

0.039

 

(0.043)

(0.047)

(0.043)

Productivity

−0.020

−0.043

−0.046

 

(0.058)

(0.053)

(0.040)

Star scientist

0.114

0.136*

0.151*

 

(0.096)

(0.080)

(0.087)

Cognitive proximity

−0.012

0.023

−0.025

 

(0.126)

(0.106)

(0.112)

Research diversity

0.133

0.104

0.208

 

(0.164)

(0.150)

(0.151)

Network proximity

0.267***

0.252***

0.338***

 

(0.041)

(0.037)

(0.051)

Size

 

−0.014

−0.088

  

(0.082)

(0.083)

Age

 

−0.488

−0.262

  

(0.382)

(0.377)

Shanghai ranking

 

−0.001

−0.023

  

(0.104)

(0.097)

Private partnership

 

−0.470***

−0.185

  

(0.163)

(0.195)

Past EUROCORES grant

 

0.074

0.049

  

(0.061)

(0.069)

Budget requested

 

0.113***

0.078*

  

(0.043)

(0.041)

Number of countries

 

0.013

0.056

  

(0.096)

(0.113)

Female ratio panel

  

−0.298

   

(0.294)

Productivity panel

  

0.005

   

(0.081)

Panel size

  

−0.353***

   

(0.100)

Age panel

  

−1.094**

   

(0.555)

Panel workload

  

−0.281***

   

(0.071)

Shanghai ranking panel

  

0.114

   

(0.113)

Programme fixed effects

 

Year dummies

Domain dummies

Observations

511

511

511

AIC

509.903

508.441

518.656

BIC

708.921

737.100

658.392

Log likelihood

−207.952

−200.221

−226.328

Deviance

415.903

400.441

452.656

  1. This table reports marginal probability effects of the probit estimation, while holding all regressors at their mean value. The dependent variable is ‘Passed 3rd Stage’ as defined in Table S1. All other explanatory variables are also defined in Table S1. The three models include different configurations of explanatory variables to test the sensitivity of the estimation. Robust standard errors in parentheses, clustered at programme-level: ***, **, *, indicate significance at the 1%, 5% and 10% level, respectively.