This paper by Robert W. Fairlie, published in November 2003, discusses an extension of the Blinder-Oaxaca decomposition technique to logit and probit models. The Blinder-Oaxaca decomposition is a widely used method to identify and quantify the contributions of group differences in measurable characteristics to racial and gender gaps in outcomes. However, this technique cannot be directly applied when the outcome is binary and the coefficients are from a logit or probit model. Fairlie proposes a method to perform a decomposition using estimates from these models, providing a more thorough discussion of how to apply the technique, analyzing the sensitivity of decomposition estimates to different parameters, and calculating standard errors. The paper begins by introducing the Blinder-Oaxaca decomposition and its application in identifying racial and gender differences in outcomes. It then presents a non-linear decomposition technique for binary outcomes, which involves using the cumulative distribution function of the logistic or normal distribution. The technique is applied to the racial gap in home computer ownership rates, with estimates from logit and probit models showing that income, education, and family characteristics are significant contributors to the gap. Fairlie also addresses issues such as the ordering of variables in the decomposition, the use of sample weights, and the comparison of results with the standard Blinder-Oaxaca decomposition. The paper concludes by highlighting the advantages of the non-linear decomposition technique, particularly its applicability to models where the dependent variable is not a linear function of the explanatory variables.This paper by Robert W. Fairlie, published in November 2003, discusses an extension of the Blinder-Oaxaca decomposition technique to logit and probit models. The Blinder-Oaxaca decomposition is a widely used method to identify and quantify the contributions of group differences in measurable characteristics to racial and gender gaps in outcomes. However, this technique cannot be directly applied when the outcome is binary and the coefficients are from a logit or probit model. Fairlie proposes a method to perform a decomposition using estimates from these models, providing a more thorough discussion of how to apply the technique, analyzing the sensitivity of decomposition estimates to different parameters, and calculating standard errors. The paper begins by introducing the Blinder-Oaxaca decomposition and its application in identifying racial and gender differences in outcomes. It then presents a non-linear decomposition technique for binary outcomes, which involves using the cumulative distribution function of the logistic or normal distribution. The technique is applied to the racial gap in home computer ownership rates, with estimates from logit and probit models showing that income, education, and family characteristics are significant contributors to the gap. Fairlie also addresses issues such as the ordering of variables in the decomposition, the use of sample weights, and the comparison of results with the standard Blinder-Oaxaca decomposition. The paper concludes by highlighting the advantages of the non-linear decomposition technique, particularly its applicability to models where the dependent variable is not a linear function of the explanatory variables.