Preliminary test, small sample size, continuous outcome — paired t test should be preferred unless there’s a specific cutoff that discipline is looking for (eg score=3 would suggest a bad quality of sleep?) then it would start making sense to dichotomise the original continuous outcome var.
Preliminary test, small sample size, continuous outcome — paired t test should be preferred unless there’s a specific cutoff that discipline is looking for (eg score=3 would suggest a bad quality of sleep?) then it would start making sense to dichotomise the original continuous outcome var.
Just my two cents though