Date of Degree


Document Type


Degree Name



Educational Psychology


Sophia Catsambis

Subject Categories

Educational Psychology | Education Policy | Pre-Elementary, Early Childhood, Kindergarten Teacher Education


charter schools, dose-response, first grade, Kindergarten, propensity scores, school autonomy


School autonomy is at the core of influential educational policies aimed at improving school effectiveness and students' academic performance both in the United States and abroad. Initiatives that promote a transfer of authority from higher levels of the school system to local schools, such as the charter school movement and School-Based Management (SBM), have become increasingly popular in the last two decades. These initiatives operate under the premise that local stakeholders (principals, teachers, and parents) understand their students' needs better than higher-level administrators, which enables them to make better educational decisions regarding students' academic success. However, despite the prominence of such decentralization efforts in the current policy environment, evidence demonstrating the effectiveness of autonomous schools on students' learning is surprisingly limited and inconsistent.

The question of primary interest in this dissertation is: What are the effects of increased school autonomy on schools' average reading gains for students in Kindergarten and first grade? Data for this investigation came from the Early Childhood Longitudinal Study - Kindergarten (ECLS-K), a nationally representative sample of schools and of students who were followed from Kindergarten through eighth grade. The independent variables were constructed from a series of questions about decision-making answered by the principal in each school, resulting in three measures of school autonomy: (1) school autonomy over instructional matters; (2) school autonomy over personnel/administrative issues; and (3) a global measure of school autonomy.

To minimize the problem of estimation bias inherent in observational studies such as this, and to account for the non-binary nature of the independent variables, propensity scores for dose-response models were used (Joffe and Rosenbaum, 1999). Using this approach, we are able to compare similar schools (those with similar propensity scores) that are exposed to different doses of school autonomy. Such a comparison allows us to estimate the effects of incremental doses of the treatment (school autonomy) on the outcomes variables (reading gains). For each measure of school autonomy, one propensity score - defined as the probability of assignment to a particular dose of the treatment, given a set of covariates - was computed per school. The effects of the treatments were estimated by regressing each outcome variable on the doses of school autonomy, controlling for the respective propensity score. This was done separately for each measure of school autonomy.

The results of these analyses show that none of the treatment effects is statistically significant. Increased doses of school autonomy are not associated with higher reading gains in Kindergarten or first grade in this nationally representative sample of schools. The conclusion was the same when I estimated the treatment effects using a subclassification method. Surprisingly, increased doses of every school autonomy measure were found to have positive effects on schools' average reading gains for a particular subgroup of schools: private schools that adopted School-Based Management (SBM). This finding, however, is based on a small number of schools and requires further investigation.