This research will analyze the factors that influence how long math and science teachers stay in teaching. These factors include personal demographic characteristics, subject area specializations, salaries, and school district characteristics. The investigators will use methodologies, known as waiting time models, that have not been used previously in studying teachers' career paths. They will apply these powerful methodologies to three longitudinal data sets, from rich data on individual teachers in Colorado, Michigan, and North Carolina. The research will emphasize comparisons between math and science teachers and other teachers, between teachers trained to teach particular sciences, between time periods (early 1970's versus early 1980's), and between regions of the country. The results should be valuable for two reasons. First, they will inform policy discussions about how to increase the attractiveness of the teaching profession to college graduates with math and science training. Second, they should provide information useful in improving the teacher demand and supply models used to predict teacher shortages and surpluses.