At the moment the TestType is
data TestType = OneTailed
I think it makes more sense to be
data TestType = LeftTailed
so that e.g. in a t-test you can specify H1 to be mu /= 0, mu < 0 or mu > 0.
mu /= 0
mu < 0
mu > 0
The workaround at the moment is to supply OneTailed to the test, and G.map negate v over the input sample.
G.map negate v
Yes, you're right on this one. However it's not only problem with current API for statistical tests (see #25 for example). It need big rewamping and I would rather do it in one go.
I'll try to summarize curent sate of statisticsal tests later.
Fixed in confidence branch.