Hi, Are interaction and correlation between factors the same thing essentially? How do we apply the different techniques in the model to account for correlation between factors? I learnt that in GLM, we can allow for complete interaction or marginal interaction to refine the model. For one-way, two-way analyses e.g. Cramer V's statistic, these are applied before GLM. Am I right?
They're not the same - see page 40 (onwards) of chapter 16. One-way tests etc are normally looked at before the full GLM, as they are only simple statistical tests.
I understand that interaction is looked at during the GLM and one-way/two-way tests are applied before the full GLM. Nevertheless, both measure correlations between the factors?
I suggest you have another read of page 40. The difference between interaction and correlation are explicitly asked as a self-assessment question there. Also, if you look through the SP8 forums, you'll find the question has been asked (and answered) at least 3 times before! Always worth searching there first - you'll also find more explanations of your questions about factors etc. p.s. worth taking a look at all the GLM questions in the exam since 2010, as pretty much all these concepts have been examined (with numerical examples) since GLMs entered the syllabus. Hope that helps! Ian
