Gender Differences at Critical Transitions in the Careers of Science, Engineering, and Mathematics Faculty (2010)

Chapter: Appendix 3-2: Estimated Adjusted Mean Effects and Differences for the Probability That There Are No Female Applicants

Previous Chapter: Appendix 3-1: Review of Literature and Research on Factors Associated with a Higher Proportion of Female Applicants
Suggested Citation: "Appendix 3-2: Estimated Adjusted Mean Effects and Differences for the Probability That There Are No Female Applicants." National Research Council. 2010. Gender Differences at Critical Transitions in the Careers of Science, Engineering, and Mathematics Faculty. Washington, DC: The National Academies Press. doi: 10.17226/12062.

Appendix 3-2
Estimated Adjusted Mean Effects and Differences for the Probability That There Are No Female Applicantsa

Differences Across Effect Levels

Estimated Mean Difference (Lower 95%, Upper 95% Confidence Limits)

Biology – Chemistryb

0.22 (–0.08, 0.51)

Biology – Mathematics

0.50 ( 0.01, 0.99)

Biology – Electrical engineering

0.23 (–0.12, 0.57)

Biology – Physics

0.22 (–0.11, 0.54)

Biology – Civil engineering

0.13 (–0.07, 0.34)

Tenured – Tenure-track

0.81 (0.71, 0.92)

Private institution – Public institution

0.66 (0.49, 0.84)

Top 10 department – Next 10 depts.

0.27 (0.10, 0.44)

Next 10 departments – Remaining depts.

0.81 (0.59, 1.03)

M – F search committee chair

0.24 (–0.16, 0.63)

a The sample size used to fit this model was 667. The effects fit were: (1) indicator variables for discipline (Biology, Chemistry, Civil Engineering, Electrical Engineering, Mathematics, and Physics, (2) indicator variables for Tenured, Tenure-track, (3) indicator variables for private institution, public institution, (4) indicator variables for top ten departments, second ten departments, and remainder, and (5) an indicator variable as to whether the committee chair was female.

b The estimated adjusted mean differences can be interpreted using Biology – Chemistry as an example. For those individuals in Biology, there is an estimated probability of having no female applicants given, or conditional on, the values for the remaining predictors in the logistic regression model. There is an analogous set of estimated conditional probabilities for Chemistry, again conditional on the predictors in the model. For each set of predictors, one can compute the difference of the estimated probabilities, and then one can average these differences in estimated probabilities over the estimated distribution of the predictors. The result is an estimated average difference of probabilities.

SOURCE: Departmental survey conducted by the Committee on Gender Differences in Careers of Science, Engineering, and Mathematics Faculty.

Next Chapter: Appendix 3-3: Estimated Adjusted Mean Effects and Differences Based on the Modeled Probability of the Percentage of Applicants That Are Female
Subscribe to Email from the National Academies
Keep up with all of the activities, publications, and events by subscribing to free updates by email.