I am conducting a one-group pre-intervention / post-intervention test. All the subjects completed the intervention as otherwise there would be ethical concerns. Is there any test other than a paired t-test that can be used to also take into account covariates, such as time, experience, etc?
1
$\begingroup$
$\endgroup$
0
$\begingroup$
$\endgroup$
There are problems with any alternative but there are also problems with the paired t in this situation.
But one thing you could do is look at whether score (or result or whatever) is related to whatever it is you are interested in and add pre- vs. post- as a variable.
One problem is that, unless the measure is close to perfect, you could be testing the error rather than the effect of the intervention.
answered Mar 7 '17 at 20:30
