Defining Academic “Quality”: The Gendered Effects of Admissions Criteria in Doctoral CS and CE Programs Holly R. Lord J. McGrath Cohoon Gender Diversity in Computing Workshop
Big Picture, Bottom Line, Overview Recruitment practices affect women’s representation Admission criteria affect women’s representation Institutional characteristics affect women’s representation
Data are from a large study Three surveys in Fall 2003 Stratified random sample of 48 departments 775 faculty members 1976 students Master’s and PhD Response rates 94% chair 63% faculty 55% grad students
Notable Gender Differences in Program Choice Women more than men emphasized Geographic preferences or constraints Impression of faculty from campus visit Department culture Flexibility in program content
Recruitment Model Regression on Women's Representation in Graduate CSE Programs Beta for all departments Beta for departments where effort >=2.6 n=46n=34 Flexibility in program content0.21*0.18 Personal recruitment by men-0.49**-0.35* Effort to enroll women0.29*0.36* Adjusted R **
How do admissions criteria affect women’s representation? Admissions criteria formal index ambiguous index other criteria Admissions-related attitudes
Dependent Variable: female proportion PhD students
Controls Public/private 1993 National Research Council rank Program size Student/faculty ratio Proportion of female faculty Carnegie classification
Ambiguous criteria are commonly used to assess applicant quality #1 General quality of academic record #2 Motivation #4 Academic letters of recommendation #6 Communication skills
Extensive & Intensive differ in emphasis Grades in computing courses85% 94% Communication skills73%64% Res. Exten.Res.Inten. % of faculty rating criteria very or extremely important
Faculty want to increase diversity Most agree their department should actively recruit underrepresented groups (80%) Few faculty believe CSE is inherently unattractive to women (21%)
Women’s Representation is Lower in Research Extensive Institutions Research extensive Female proportion PhDs -
Formal index exhibited no effect Formal index (functionally relevant, specific) grades in computing courses math background GRE Score reputation of undergraduate institution/program No measurable relationship with women’s representation
Ambiguous criteria has a positive affect Ambiguity Index general quality of academic record motivation communication skills maturity academic letters of recommendation women’s representation +
Life experience criterion favors women Consider life experiences Ambiguous criteria Female Proportion of PhDs + + +
Diversity as a criterion favors women Membership in an underrepresented group
Gendered criteria has an affect Computing work/volunteer experience negatively affects the gender balance
Final Model Regression Results for Women's Representation Betat test Sig. Carnegie Classification Member of underrepresented group Computing work or volunteer experience Consider life experience Constantns Adjusted R-squared for model
Questions?