I have been told that there are two ways to unstandardise the path coefficients.
The first is to use the Smartpls output the LV scores unstandardised and run a simple linear regression
The second is to use Smartpls output outer weights (ow) for each LV and then calculate
Βetaunstandardised = (∑ owe/se / ∑ owx/sx) * Betastandardised (PLS)
owe , se are outer weights and standardev. (from the orginal data) for the endogenous LV
owx , sx are outer weights and std. for the exogenous LV
Which one should a use, I get two different solutions? Can somebody clear things out for me?