A new Youtube video is available for WarpPLS:
This video shows how one can conduct an SEM analysis using WarpPLS, save that analysis with a different project name, change the resampling method (from bootstrapping to jackknifing), and then redo the analysis.
At the end, the user has two project files, one with all of the P values calculated through bootstrapping, and the other with all of the P values calculated through jackknifing.
As noted in the WarpPLS User Manual, bootstrapping and jackknifing provide a good complement to each other in the context of warped PLS-based SEM.
Thus, some users may want to run two analyses of the same model, one with each resampling method, and use the results that are associated with the most stable resample path coefficients. These will typically be the ones with the lowest P values, since P values go up as the standard errors in the resample set go up. High resample standard errors are associated with instability. The instability itself often comes from outliers, which may drastically change the shape of a warped relationship in each resample.
Well, moving from statspeach to plain English, there are good theoretical reasons to recommend that users choose the most stable results (i.e., with the lowest P values) as the results that they will use in research reports, whether they are obtained with bootstrapping or jackknifing. The choice may be made individually, for each path coefficient. This should be disclosed to the readers of the report; a sentence like this would probably be enough: "Both bootstrapping or jackknifing were used in the analyses. The results reported here are those associated with the most stable resample estimates."