At the moment, the ImageStatistics view is unable to calculate statistics for an input image and a binary mask if the number of time steps for those two images differs.
However, a common use case is having a mask with only one time step that you want to apply to each time step of the input image.
A better solution than prohibiting the calculation if the time steps differ, would be to always calculate over the input image's current time step with that same time step of the mask image. If the mask does not have that time step, its last available time step should be used instead. Furthermore, the used mask time step should be displayed so the user knows what's happening.
Also, right now the statistics are not recalculated when the time step changes (time slider in image navigator).