Goodness of Fit Measures for Generalized Estimating Equations. An Empirical Evidence on Relationships Between Gender Diversity and Corporate Performance

The principal idea behind generalized estimating equations is to generalize and extend the usual likelihood score equation for a generalized linear model by including the covariance matrix of the clustered responses. The advantage of the GEE is that we do not need to specify the entire response distribution. Only the mean structure, the mean-variance relationship and specification of the covariance structure need to be defined. In order to evaluate the strength of association between a response variable and covariates, coefficient of determination based on Wald Statistics, Likelihood Ratio and pseudo-R-2 based on a quasi-likelihood method are presented and some relationships among them are shown. Following the strand of literature that examines the relationship between gender diversity and corporate performance, this study aims to analyze whether and how the presence of women in corporate governance bodies affects the corporate performance (ROA) of companies belonging to the European banking sector.