I know you can generate the covariance matrix by entering covariance values and using cov instead of corr when entering them. However, I haven't quite found a way to input it to run the regression. When I find something on this, I'll certainly post it! But keep in mind that since I was adding in standard deviations along with the correlations, the model generates the unstandardized coefficients as though it was being run using covariances. Remember that Pearson's r = cov(x,y)/(sd(x)sd(y)) , where cov is covariance and sd(x) and sd(y) are standard deviations for x and y. From this, you can see that by multiplying each Pearson's r by the product of the standard deviations for two variables, you can reconstitute the covariance matrix. Hope this helps! Cheers.