Each ? is interpreted as the change in log odds of a success for a 1-unit increase in the corresponding predictor, holding all other variables constant.
If the predictor is binary, it represents the change in log odds for that setting of the variable compared to its reference level. Exponentiating the coefficients converts the interpretation to the change in odds rather than log odds, which is easier to conceptualize. If interaction or polynomial terms are present in the model, the effect of the coefficient can change depending on the value of X, which must be dealt with appropriately.
