It seems that you get different Bayes Factor values for ttestBF() and generalTestBF() when using a data set with an unused factor level whose underlying integer value is in between two of the actual used factor levels. I think this discrepancy is ultimately due to the fact that generalTestBF() calls lmBF(), which calls reFactorData() on the data set; however, this does not happen for ttestBF().
It seems that you get different Bayes Factor values for
ttestBF()
andgeneralTestBF()
when using a data set with an unused factor level whose underlying integer value is in between two of the actual used factor levels. I think this discrepancy is ultimately due to the fact thatgeneralTestBF()
callslmBF()
, which callsreFactorData()
on the data set; however, this does not happen forttestBF()
.Here's a way to reproduce this:
Note that
group
is a factor with three levels, since it hasn't been releveled after dropping "b".Now, when you call
ttestBF()
vs.generalTestBF()
, you get:0.1833829 != 0.2485882 👎
Compare this to when you drop "c" instead of "b":
You still have an unused factor level, but now you get equivalent results with
ttestBF()
andgeneralTestBF()
:0.2177451 == 0.2177451 👍