comparison of path coefficients

[sandra]

Hi,

I would like to test if one path coefficient (e.g. b1) is bigger than another (b2). Usually I would use (after Regression):

t=b1-b2/(varb1+varb2-2covb1b2)^0,5

Is this test possible with PLS coefficients, too? Unfortunately the covariance is not provided - or does PLS assume the Cov to be zero?

Thanks a lot

Sandra

[stefanbehrens]

Hi Sandra,

given that - at a structural level - PLS is just a multiple regression between latent variables, your test should work.

While the covariance of coefficients is not reported by SmartPLS, you could simply export the path coefficient data from the bootstrap report to Excel and then calculate the covariance there.

Cheers,
Stefan

[Diogenes]

Hi all,

a reference in viewtopic.php?t=172

to create an I.C. with bootstrapping, since PLS doesn´t have normality assumption.

Best regards