Significance affected by number of cases in bootstrap

moataz · Post by **moataz** » Wed Apr 11, 2012 3:13 pm

Dear All,

I have a sample of 643 observations. Upon testing my model, some paths in the model were significant. However, when I split my sample into two sub-samples (for group comparison based on gender) I found that at least one -previously significant - path was no more significant for both sub-samples. Please note that, when running bootstrap for sub-samples, I have used the number of cases for the respective sub-sample, rather than the original sample. Specifically, 301 cases for males and 342 for females. Now, when I re-run the bootstrap algorithm for sub-samples, but using the original sample's number of cases (643), I get a more meaningful result.

Sorry for the long introduction, but now the core question is: the number of cases in bootstrapping must be equal to the number of cases in the sub-sample or can be equal to or more than that number?

Thank you!

Moataz Soliman

Hengkov · Post by **Hengkov** » Wed Apr 11, 2012 7:49 pm

Hi Moataz,
For Multigroup Analysis, number of cases equal to sub-samples (male or female).
See some literature below:
Chin, W.W. 2000. FAQ PLS http://disc-nt.cba.uh.edu/chin/plsfac.htm
Keil et al. 2000. A cross-cultural study on escalation of commitment behavior in software projects. MIS Quarterly (24:2), pp.299-325
Henseler, J. 2012. PLS-MGA: A Non-Parametric Approach
Sarstedt et al. 2011. Multigoup Analysis in PLS Path Modeling. Advances in International Marketing (22), pp.195-218.
Regards,
Hengky

moataz · Post by **moataz** » Wed Apr 11, 2012 8:00 pm

Hi Sarjana,

Thank you very much for your prompt and informative reply.

Kind Regards,

Moataz

Hengkov · Post by **Hengkov** » Wed Apr 11, 2012 8:09 pm

Ok, your welcome and good luck.
Regards,
Hengky