We understand the importance of having qualified, effective teachers in every classroom. We have learned from many research studies, particularly those of William Sanders and his colleagues in Tennessee, that students who are taught by effective teachers (defined by Sanders as those whose students consistently post gains in student achievement scores) for several years in a row will experience the benefits throughout the rest of their school careers and beyond. After three years with the most effective teachers, students show achievement gains significantly higher than those of students with the least effective teachers.
We can reasonably hypothesize that more experienced teachers will exceed the effectiveness of recently inducted beginning teachers. Further, as is now widely recognized in most states, new teachers need and benefit from support during their induction period. Support during the new teachers' first year or two may be just as important to their effectiveness as their pre-service training, their state certification, and their subject matter skills. To justify assigning resources to provide support for novice teachers, legislators and school district administrators need to be convinced that such support is associated with educational outcomes beyond participant satisfaction. Researchers have shown that induction and mentoring programs may have a positive effect on teacher retention. However, few studies demonstrate any connection between new teacher induction and student achievement, the outcome that is probably of most interest to parents, educators, and legislators. Perhaps the main reason for this is that such studies are diffi cult to conduct. First, it is hard to obtain the necessary data. Many schools and districts do not maintain databases connecting student test scores to teachers. Many states do not test students in all grade levels annually, and tests are changed frequently, making it diffi cult to compare performance from year to year. Also, induction programs vary, and many factors contribute to changes in student achievement besides the kinds of support beginning teachers receive. These include school variables, family, economic status, and social issues; other kinds of support such as teacher aides, subject-matter specialists, tutoring; teaching to the test; language issues; and students' health and mood at the time of the testing. Finally, not all educators agree on the validity of using standardized test scores to measure student learning.
Imposing an experimental design on treatment and subjects would address all of these issues, except the last. However, the most challenging aspect of this field is often securing access to a suitable control or comparison group of any sort, much less one meeting the standards of an experimental design. These dilemmas force compromises that can make interpretation more difficult.