I have 15 dependent variables on on IQ measure am using in my study but after checking for skewness and kurtosis two of them are not normally distributed.How can I justify this and go ahead to use parametric tests for my analysis
Which parametric test are you using? For regression and ANOVA for example the residuals not the individual variables have to be normal. This is a point that a lot of text miss when they discuss normality (I would be rich if I had a dollar for every poster here who pointed this out to me).
Not that rich. There have only been a few of us that point it out. We've had to do it MANY times but there are only a few members that have pointed that fact out to you
Lol yeah I mispecified my statment. I should have said times not posters
To the OP normality will not bias the results. It will distort confidence intervals although I believe most parametrics are very robust to the assumptions of normality. For example ANOVA.
