Parameters in GLM are not interpreted as in ordinary least squares. That is, a unit change in an independent variable k does not correspond to a bk change in the dependent variable. This is because GLM uses a nonlinear link function. If coefficients are examined, it is in terms of first differences: the researcher determines two levels of interest for a given independent variable, then calculates the value of the dependent variable under these two conditions, holding all other variables constant at their mean value.