Leslie 9311531 Underrepresentation of minorities and women in science and engineering fields is a matter of major national interest and social concern. The purpose of the study is to identify and quantify, through econometric analysis, variables explaining majors to graduation, and entrance into graduate S/E programs and S/E careers. The primary focus will be on females and minorities, but majority males will be examined as well. Previous research has efforts for minorities and women have been sparse and limited in scope. A full array of variables will be examined including market factors. Data for the study will be extracted primarily from the National Longitudinal Study of the High School Class of 1972 (NLS72), High School and Beyond (HS&B), and the longitudinal studies of the UCLA Cooperative Institutional Research Program (CIRP), which will provide data for the S/E undergraduate choice and persistence analyses and persistence analyses and some data for the S/E graduate program and career choice analyses. Other data bases for use in S/E graduate program and career choice analyses may include NSF's, The Experienced Sample of Scientists and Engineers, the Survey of Recent Scientists and Engineers, the Survey of Doctorate Recipients, and the Scholastic Aptitude and Graduate Record Files. Analytical techniques include, for selection of major or occupational choice, determining the probabilities of various choices for distinct groups, e.g., females or males, and then determining the choices other groups would make based on specified probabilities. The gender differences in the distribution among choices reflect gender differences in the explanatory variables. Multinomial logit or, alternately, a conditional logit model is used to analyze choice of college S/E major. Multinomial probit is also used if independence assumptions are not satisfied. For analyzing persistence, Markov or multinomial logit choice model, is used. Early job history is modeled as a simple logit choice model, while choice of S/E graduate program will be analyzed using a basic choice framework or multinomial choice framework. Choice of given model or analytical technique will be dependent on satisfying the assumptions of the model or technique.
