Four essays on the influence of school ecologies on educational production

Having a better understanding of what factors are most influential in enhancing student achievement is crucial for effective education policy-making. This dissertation, consisting of four essays, seeks to make a contribution to the educational production literature by conducting an in-depth empirical investigation of the role of various school ecological factors in educational production using an expanded vector of student outcomes including math and reading tests, GPA, effort hours, and absences. First, the relationship between exposure to school violence and student outcomes is examined by expanding the traditional vector of inputs to include variables accounting for a student's direct exposure to school violence, the level of fear amongst students, school disciplinary structure, neighborhood safety levels, and the student's troublemaker status. The findings suggest that exposure to school violence has an impact on student outcomes similar in magnitude to more commonly investigated inputs such as parental education and student/teacher ratio. Second, a study of peer effects is conducted in a novel way through the introduction of the peer friction coefficient, representing a measure of the level of difference between the characteristics of the individual student and the average peer. Using this variable, two hypotheses are tested, which posit that the greater the level of peer friction, the less likely the student's needs are going to be met by teachers and the weaker peer effects will be. Empirical support is provided for both of these hypotheses. Third, the impact of diversity within the student population on student outcomes is explored through the innovative application of ecological diversity measures to the school environment. Specifically, Shannon's entropy is used to derive measures of racial, gender, and ability level diversity that are used as inputs into the educational production function. These measures are shown to be influential, having a differential impact between high and low achieving students. Finally, a comprehensive ecological model of educational production is derived, integrating the models of the three earlier studies, in order test for the simultaneous significance and any potential interaction between the key variables of interest. Support is found for all key variables and an extensive network of significant interactions.