METHODOLOGICAL ISSUES IN THE DESIGN OF LONGITUDINAL RESEARCH - PRINCIPLES AND RECOMMENDATIONS FOR A QUANTITATIVE STUDY OF TEACHERS CAREERS

In this paper, we initiate a discussion of the possible methodological features of a potential new national longitudinal study of teachers' careers. We use a review of the substantive and methodological literat ures and ''pseudolongitudinal'' analyses of data drawn from the Nation al Center for Education Statistics' Schools and Staffing Survey and Te acher Follow-Up Survey to support our arguments. Our presentation is f ramed by six key principles of research design that are then used to s upport concrete recommendations about whom should be studied, how ofte n they should be observed, and for how long the study should continue. Our six design principles assert that, in the new study we must colle ct truly longitudinal data, view time as both an outcome and a predict or collect data on both time-varying and time-invariant measures, coll ect data prospectively whenever possible, collect data beginning in mu ltiple base years, and collect data at all relevant levels of the orga nizational hierarchy. Although it is impossible to define a single res earch design that is optimal for answering all research questions abou t the teaching career we recommend that the new study should sample fr om the population of teachers who are beginning their first, second, t hird, and subsequent spells, that each of these teachers be followed f or at least 12 years (both in and out of teaching), and that they shou ld be measured on at least six occasions during this period. We also r ecommend that the study be replicated starting in two, if not three, b ase years. We welcome public comment and discussion of our proposals.