One Tailed or Two Tailed MGA

[Taolin]

Hi everybody

I am currently writing a paper on changing effects over time and I am using PLS-MGA to investigate whether differences between the models (timepoints) are significant.

I am now wondering about which t-test to use. In the introductional video (https://www.youtube.com/watch?v=qssDqR_-Q7Y), the two-tailed test is used, but as far as I understand the goal is to see whether there are "differences".

In our paper, we do not hypothesize about "differences" but we hypothesize that an effect will "increase" or "decrease" over time. So from my understanding, I should use a one-tailed t-test, is this correct?

Furthermore, I would then only look at values of .9 and higher, and values of 0.1 and lower - depending on the hypothesis - as being significant (5%), is this correct?

Best and thank you in advance,
Matthias

[Hengkov]

Keep using two-tailed test. You only need to see the results of PLS-MGA on the approach used.