UCLA

A fundamental goal in education is to provide access to quality education and educational opportunities for every student. Classroom processes, teaching, and students’ learning experiences are at the heart of quality education and, as such, must be the key focus in investigating the issues of equity in access to education (O’Sullivan, 2006; Peske & Haycock, 2006; Raudenbush & Sadoff, 2008). In light of India’s performance in PISA 2009, this dissertation study investigates the larger issues of access to high quality teaching practices, and other valuable school resources to get a better picture of India’s poor performance. To this end, publicly available large-scale datasets such as the Programme of International Student Achievement (PISA; by OECD), and the Teaching and Learning International Survey (TALIS; by OECD) enable us to look beyond student’s achievement scores or a country’s ranking by providing us with a plethora of information on students, teachers, and schools. Moreover, even though PISA assessments are low stakes tests, they often drive high stakes education policy decisions in multiple countries.

In this dissertation study, I make use of the PISA student and school questionnaires for India along with state-of the art multilevel IRT models implemented using MCMC. I describe and illustrate a methodology to examine students’ exposure to key instructional practices based on students’ responses to PISA survey items, and then use this measure as an outcome variable in a three-level IRT model to investigate differences in the amounts of exposure to key practices within schools and between schools. Measurement invariance was established across the rural and urban regions of Himachal Pradesh (HP) and Tamil Nadu (TN) before comparing the construct of interest across various sub-groups. This set of analyses indicates that the items in the student questionnaires capture the construct of interest, and are not an artifact of underlying translation errors, or cultural differences in the examinees understanding of these items.

A multilevel IRT approach, such as the one employed in this dissertation allows us to tease apart the variation in the extent to which students experience particular instructional practices into their within-school and between-school components. The analysis strategies developed in connection with my dissertation will hopefully be valuable to other researchers interested in investigating questions concerning inequality in the distribution of key instructional practices.

Lastly, in chapter 6, I depict the use of this approach to identify schools, whose students on an average, experience relatively high or low exposure to the instructional practices of interest. Futhermore, a key finding of this set of analyses indicated that the that a majority of the public or government-run schools were concentrated in the lowest end of the socio-economic scale; private schools were found to be more spread out, but still in low socio-economic areas.