Hi Sneaks,

although you are obviously right to stand for the best method, I think you may be on slippery grounds for your viva. So just to play devil's advocate and as a little exercise, why don't you explain in a few words what your method in about, why it is better than a student t-test and as such why many of the established researchers in your field use the "wrong" method?

:-)

The reasons are well documented for anyone that wants to look. The frequentist GLM approach (e.g. ANOVAs, t-tests) are based on the normal distribution - the accuracy of the test depends on your data fitting this model distribution; however real life data often doesn't. I know we all know that data have to be normally distributed to use these tests, but the fact is a lot of social scientists just use them anyway even when the assumptions are not met. The benefit of bootstrapping is that you use your sample as the reference for your statistic i.e. you don't make these assumptions about the distribution so the results are more robust. The reason a lot of behavioural scientists don't use bootstrap is simply because it takes longer - you end up using MATLAB or something, and you have to learn the language, plus the bootstrapping itself takes time to run. Compare that to the ease of plugging your data into SPSS or Statistica, and it spitting out a nice p-value... people just don't want to do it because they didn't get into this field to end up doing math, and if the academics are happy with the status quo they just say to their students "oh don't bother with that, just do a t-test". Then the students become the next generation of academics, and they say the same thing to their students.

thanks Melsie, I might steal that explanation haha. Andy field has started to do a bit on bootstrapping, because the newer versions of SPSS have a button alongside most of their tests saying 'bootstrap' - there's a podcast on his website about it which is really easy to follow for anyone who wants to know more.

One reviewer comment is that they've said "I understand the principle of bootstrapping, but don't know why you've applied it here" - I've actually explained that I've applied it because it adds robustness etc. I don't really understand why if you were a reviewer you'd raise the issue if you clearly don't understand its use (??)