The decomposition of inter-group differences in a logit model: extending the Oaxaca-Blinder approach with an application to school enrolment in India

This paper suggests a method of decomposing differences in inter-group probabilities from a logit model and shows how it can be related to similar decompositions derived from a Oaxaca-Blinder framework. In so doing, it offers a solution to a problem, embedded within the Oaxaca-Blinder decomposition, relating to the appropriate choice of common coefficient vectors with which to evaluate the different attribute vectors. The decomposition method also shows how pair-wise comparisons of groups might be conducted in the presence of more than two groups, without discarding the information on groups excluded from the comparison. The proposed method is applied to inter-group differences in schooling participation in India and the results are compared with the Oaxaca-Blinder method. The decomposition is applied specifically to inter-group differences in the enrolment of boys at school in India. Keywords: Logit regression, school enrolment, India, decomposition methods