Measuring homophily in the choice of upper secondary education: A new socio-metric instrument

Abstract

The aim of this thesis is to advance the methodology of homophily research, by providing a new socio-metric instrument that can be used by a novel methodology developed by Jansson et al. (under publishing) (Jansson 2017). A socio-metric instrument is a tool which provide information of individuals’ relationship with others. The choice so of socio-metric instrument implies specifying the relational content one is interest in and linking socio-metric instruments to actions is not always straight forward. The socio-metric instrument I propose in this thesis is based on lower secondary school graduates’ choice of upper secondary education. Hence, the socio-metric instrument is especially useful for researchers interested in segregation in lower and upper secondary education. It can be used to test what background dimensions are important for which classmates’ the lower secondary graduates gravitate towards, in their choice of upper secondary school. This have direct implication on the demographics of upper secondary schools and can have implication on the informal segregation in the destination upper secondary school. It can also be used to approximate self-selection bias for researchers interested in estimating peer effects in lower secondary school. This choice has two components; track and school. I use findings from previous research to optimize the social considerations of the graduates in this choice, which I operationalize into the socio-metric instrument that measures friendly coordination of choices. In summary I expect graduate friends that coordinate their choices to apply more often to the same upper secondary school. Furthermore, I expect friends to do, to a far less degree than non-friends or non-coordinating graduates, is to apply to the same specialization track but at different schools. The socio-metric instrument is used in empirical analyses of gender and ethnic/ immigrant status homophily. Coupling the socio-metric with the methodology of Jansson et al. (under publishing) (Jansson 2017) shows promising ability to infer homophily from registry data. The analyses show homophily on the dimensions gender and immigrant status, and that homophily is more persistent from first to second-generation versus natives among female graduates than among male graduates. An analysis of immigrant country of origin and show trends of less homophily between natives and first-generation Somali immigrants than between natives and second-generation Somali immigrants. The thesis also provides some notions on the application of the Pearson’s correlation coefficient in the methodology developed by Jansson et al. (under publishing) (Jansson 2017). This involve that the coefficients maximum positive value is constrained by the number possible choices is smaller than the number of independent categories. And that, coefficient’s the negative maximum is constrained by the sizes of the independent categories.