inconsistency in tstats

i ran a model on some data (2 regressors, 1 is grandmean the other is a [1 -1] contrast). There are two ways of getting tstats from the model object: model.tstats and model.get_tstats(). The values given by these are different and i'm not sure that they should be given self.tstats runs the get_tstats() function in the fit.py script? Not sure which of these is best to use as they are different!