Hi,
I was wondering if there is a preference with regards to standardizing a phenotypic (continuous) variable when running a classical twin design and sex-limited twin design? Does the variable need to be standardized to work with the scripts and for best interpretation?
Thanks!
Hi Leora,
There is no need to standardize your continuous variable prior to model fitting; the scripts work with the starting values you provide for the means and the variance components. If you’re estimating three variance components, divide the phenotypic variance by 3 for their starting values; or use a somewhat higher starting value for the VE component compared to the VC, VC or VD components. Please note that it’s assumed that the variables are normally distributed, thus a transformation might be warranted if the distributions of your variables are very skewed.
Hermine.