Overly positive bootstrapping results on first run

[Howard Adams]

Hi

I have noticed that after altering my model and then running a bootstrap (50 cases, 1,000 samples), on the first run the coefficient pathway significance levels are more postive than with subsequent runs, i.e., out of 11 pathways, 5 were significant > 1.960, but on several subsequent bootstrap runs with no changes made, only 2 of the 11 pathways reported as significant.

I also noticed this behaviour on another model of 262 cases and 1,000 samples, but did not register it as a pattern until now.

I have submitted my model for examination to you.

Regards
Howard