forum.smartpls.com

Hi,

Can anyone tell me how to work out the critical values for t which is generated from bootstrapping? I want to report the statistical significance of path loadings but the published literature isn't really explicit on this. Some authors seem to use N-1 for degrees of freedom in relation to students-t distribution. My sample size was 85 so is df=84 the appropriate setting (Critical t for p<0.05 = 1.66)?

Thanks

Alistair,

your degrees of freedom (df) for the student distribution are not derived from your sample size, but rather from the number of bootstrap runs. E.g., if you ran the bootstrap with 500 runs, your df would be 499.

To determine the proper critical t-value, obviously you also need to consider whether you are running a one- or a two-sided test. The literature is generally not very explicit on this issue either.

Cheers,
Stefan

I also have a question. Are t-values available only with bootstrap function, and not with general model?

For t-values see the aswers in viewtopic.php?t=151

and since normality isn´t a assumption in PLS, the t-values are computed from bootstrapping (the answer is yes).

Best Regards

Bido

Should we take one sided or two sided significance values for the boot strap? The cut off t-values change for one sided and two sided significances.