To solve the twin problems of a projected shortfall of science workers and general science illiteracy in the United States, many educators have proposed a massive restructuring of the curriculum and pedagogy of elementary and secondary school science. This resarch suggests that science eudcators focus on such issues as course design, teaching and curriculum as well as on recruitment, rewards, and opportunities in science. The goal would be to attract that group of able students who can do science, but select other options dubbed the "second tier." By getting to know these students and finding ways to reverse their migration from science to other disciplines, it should be possible to stem the massive loss of potential science workers that occurs during the college years. To explain reasons for this "second tier" students, six nonscience graduate students and one professor were paid to seriously audit for one semester an introductory physics or chemistry course. Each participant kept a record of his or her reactions to the course (e.g., comments about an instructor's approach, content of course, etc.). To determine if these accounts were typical, data from the "Concentration Choice Study 1978-1983" of 300 Harvard-Radcliffe students were studied. The study focused particularly on male-female differences and science-nonscience predictors. Analysis of the interview data revealed that for both men and women, enjoying a science course more than all other freshmen courses was a significant predictor of their decision to major in science. A final synthesis of the research findings of these studies is provided. (CW)