 |
 |
Hi Ken Regarding thee significance, if your data set is large (as you say it is) whether p=.134 is significant is a judgement call from you, and possibly might be helpful to talk to your supervisor regarding this. Technically no, but there maybe some room for manoeuvring in this case. Do your mean/medians look significant? If so this would give you more stance in making such a judgement call. I'm sorry I don't know enough about CLT to answer your other question - maybe someone else on the list can? Thanks Ben On 12 July 2011 12:46, Kenneth Concannon <[log in to unmask]> wrote: > Hi Ben, > > Thank you for your reply, it's much appreciated. > > Re: my question on moderation, given that a relationship between an independent variable and a dependent variable is generally required to have a significance of p<.05, and since p<.05 is an conventionally agreed value chosen to ascertain significance, does this not mean that any p value greater than .05 (e.g., p=.134) is, in reality, just less significant?Would this not mean that there possibly still may exist a relationship, albeit not on a conventionally agreed "significant" level? If this is the case, could moderation not still be tested for (perhaps I'm way off the mark here, apologies if this is the case)? > > Also, my sample size is N=144. Given that it is above 100, could I use the Central Limit Theorem as justification not to transform my skewed variable? > > Thanks and best regards, > Ken > > -- Many thanks Ben Haysom-Newport BSc (Hons), MBPsS, MSc