Transforming z to t values (GLM, different error distributions)


I am comparing behavioural data collected at two populations. Generalised linear models were used with a behavioural trait as a response variable. The problem is that due to the error distributions, different families were specified for each dataset: Gaussian and Binomial family functions were used.
:confused: The question then is: is it possible to transform the z-statistic to a t and then be able to compare the results from the two populations? if so, how?
If someone can point out how to do this I'd really appreciate it. Thanks in advance.