I know 'from somewhere' that the Binomial test is a 'very conservative test', for instance testing whether or not the number of correct responses in a forced choice response task is below or above chance level. Is there any reference or study supporting that?
I think this is not correct as is, it is not "very conservative" nor even "conservative", so I suspect you won't find any references out there supporting this. The only thing one could call simply a binomial test is an exact test, based on the binomial distribution, and hence it is not conservative - it's exact. Maybe this is not a very rigorous definition, but I would say a conservative test is one that tends to find an effect less often than alpha% of the times for an alpha% nominal level (if there is no effect), but when it finds an effect that's (a more often than would be for a non-conservative test) usually a true effect.
What exactly do you mean by the word 'conservative' here? Not having enough flexibility? In that case, yes, it is so! However, you do not really have any choice.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics