Theses Doctoral

Educational Choice and Educational Space

Thomson, Kathleen Sonia

This dissertation entitled “Educational choice and educational space” aims to explore the confluence of constructed space and geographic space using a supply-side context for New Zealand’s public school system of quasi-open enrollment. In Part I, New Zealand’s state and state-integrated school system across four urban areas is analyzed spatially and analytically in an attempt to learn about supply-side motivations of individual schools for selecting students. Since 1999 there has been a gradual encroachment on the open enrollment initiative due to excess demand for certain schools altering the landscape of choice. Most studies of school choice examine household motivations to choose schools- what is referred to as the demand-side dynamic of the education market. The contribution of this study to educational choice literature is provided by the opportunity presented by New Zealand’s public education system to examine the supply-side dynamic. School motivation for choosing students is warranted by the fact that individual schools are funded on a per-pupil basis and they are able make operational decisions that include defining their own catchment areas (home zones) when oversubscribed. In order to test the hypothesis that schools are behaving selectively, I leverage boundary discontinuity design (BDD) (Bayer, Ferreira, & McMillan, 2007) and meta-analysis techniques and use census data that reflects neighborhood composition closely linked to the time at which the home zone was drawn. Household characteristics as represented by 2006 census data are within 5 years old at the time the home zone was drawn in 53-percent of cases, and range to a maximum of 7 years for the full sample of schools used in the analysis. The result is a truly unique opportunity to examine evidence of school selective behavior while accounting for logistical, geographic and market features. Across a sample of 886 publicly funded state and state-integrated schools in Auckland, Wellington, Christchurch and Dunedin I evaluate 372 schools that implemented enrollment schemes with geographically defined home zones as of the 2009 school year. I find that schools are engaging in selective behavior across one or more household socioeconomic or demographic characteristics – with indicators of deprivation and minority status consistently maintaining significance in boundary and zone discontinuity. Between 37 and 52-percent of schools sample-wide are estimated to be participating in gerrymandering in small to large ways (discontinuities were greater than 0.2 standard deviations, favoring households with affluence and non-minority status). Cases of medium to large discrimination (standardized mean differences of >0.5) are evident in 12 to 26-percent of schools. Schools are zoning out households that contain higher proportions of minorities with an emphasis on non-indigenous (Māori) minorities, in particular the Pasifika group. When examining schools separately by city, sector, gender and school student body socioeconomic status I find heterogeneity in school behavior. Findings corroborate and expand upon previous work regarding New Zealand’s policies of enrollment schemes and their adverse effects.

Part II expands the implications of school-level selective behaviors to the macro setting of the metropolitan area education market. Because school zones are not mandated to be mutually exclusive or completely exhaustive of the metro area they serve, it is possible that home zones overlap in some areas and provide no coverage in others. School choice is modeled using an ordered probit approach where the number of home zones claiming the household (census meshblock) is the dependent variable. I also calculate meshblock-level schooling opportunity sets (SOS) for the primary and secondary sectors using methods developed by (Gibson & Boe-Gibson, 2014). The SOS factors household-school proximity with school performance into a summary estimate of educational opportunity for each meshblock. The determinants of each meshblock’s choice set and SOS are modeled as a function of household characteristics, with controls for geographic and market influences. I find evidence that both the size and quality of household educational opportunities are statistically significantly associated with socioeconomic status and racial composition. Affluence is a strong predictor of additional choices and higher quality school access while racial composition continues to be negatively associated with both. The “lay of the education landscape” is then considered for each metro area using the Index of Dissimilarity and mapping techniques. The exercise reveals helpful insight into each of the four education markets. The dissertation concludes with a discussion on the implications and relevance of this work to educational reform discourse and planning for both New Zealand and abroad.

Geographic Areas


  • thumnail for Thomson_columbia_0054D_13149.pdf Thomson_columbia_0054D_13149.pdf application/pdf 9.78 MB Download File

More About This Work

Academic Units
Politics and Education
Thesis Advisors
Henig, Jeffrey
Ph.D., Columbia University
Published Here
February 5, 2016