This dissertation examines the influence of student, teacher, and school variables on English and math teachers' expectations for their students. Findings from multilevel-model analyses of data from the Education Longitudinal Study (ELS: 2002) show that student achievement and stigmatized status are the strongest predictors of teacher expectations. In this study, students could be stigmatized on any of three factors (ethnicity, SES, and native language). Consistent with previous research, teachers' expectations were predicted by student achievement (test scores in reading and math). Teachers' expectations for future student academic attainment were higher for students with higher achievement than for lower-achieving students. This lends support to the argument that teachers are generally accurate in forming expectations. However, also consistent with prior research, student stigmatized status predicted expectations, with stigmatized students receiving lower expectations than non-stigmatized students. Multiple stigmatizations were powerful--students stigmatized on all three factors, in particular, received the lowest expectations. Further analyses indicated that student achievement interacts with student stigmatization for English teachers. At low levels of achievement, teachers held equally low expectations for stigmatized (on three factors) and non-stigmatized students. But, for high levels of achievement, teachers had lower expectations for stigmatized students with equally high achievement. Specifically, stigmatized students (on three factors) received expectations that were a half of a standard deviation lower than non-stigmatized students. Teacher ethnicity also appeared to influence teacher expectations via an interaction between teacher ethnicity and student stigmatization (on three factors). For English teachers, expectations were equally high for (a) ethnically stigmatized teachers rating stigmatized students, (b) stigmatized teachers rating non-stigmatized students, and (c) non-stigmatized teachers rating non-stigmatized students. The lowest expectations came from non-stigmatized teachers (White and Asian) rating stigmatized students. These findings (which control for student achievement) suggest bias in expectations that non-stigmatized teachers have for stigmatized students. School level variables in this study (e.g., percent of students receiving free/reduced lunch, percent of students who fail the competency test on first attempt), did not have a large effect on teacher expectations. Implications are discussed regarding multiply stigmatized students, self-fulfilling prophecy, equal access to educational opportunities, and recruitment of ethnic minority teachers. / text