I am testing many things in my investigation and I could also have a side hypothesis about certain relationship. However previous research shows that there is no relationship. Does it make sense to hypothesise no effect? My sample is quite small so it is very likely that I would find support for the "no effect".
Asked
Active
Viewed 32 times
0
-
1So called equivalence testing can be usefull: https://en.wikipedia.org/wiki/Equivalence_test – Daniel Dostal May 26 '18 at 10:40
-
Look at Blackwelder's article on "reversing the null and alternative hypotheses:. – Michael R. Chernick May 26 '18 at 14:25