It seems you have decided to stati transform variables that had significant levels of skewness using a reflect and square root as a way of rectifying non-normality. First can you explain how this transformation works ? Second, can you provide a sound argument as to why it is appropriate?