Page 1 of 1
t-values
Posted: Fri Dec 13, 2013 3:42 pm
by sunilsahadev
In PLS why do we get positive t-values even when the path coefficient is negative?
Re: t-values
Posted: Mon Dec 23, 2013 1:31 pm
by nayabeyes
sunilsahadev wrote:In PLS why do we get positive t-values even when the path coefficient is negative?
Sarjana Hengky Latan reply is more perfect.
Posted: Mon Dec 23, 2013 10:06 pm
by Hengkov
Hi,
This is a bug software. Beta/standard error = T-statistic. So, beta and T-statistic must have same sign, cannot different (if P-value always positive, but T-statistic not). I already talk it two years ago, we hope in SmartPLS 3.0 already repair.
Best regards,
Re: t-values
Posted: Mon Sep 30, 2019 7:49 pm
by Nipuni Jayasundara
I also have the same problem. Positive t-value but negative path coefficient. How do I solve this issue?
Re: t-values
Posted: Sun Oct 06, 2019 3:45 pm
by admin
This is not a bug. We report the absolute t-value as indicated by the "|" around the statistic. Researchers are usually only interested in the magnitude of the t-value for significance testing. The sign of the coefficient of interest gives you the sign of your effect.