>I'd guess that the actual uncertainty >in the total is more a function of the >uncertainty in the numbers than it is in >the sampling.

totally agree. As with any indicator there is quite a bit of uncertainty. And as you suggest "there are the hires that happen without a listing..." Most career counselors would agree that there are actually more hidden jobs than what the boards suggest...

So... your argument might just boost the number if we add some multiplier into the model based on the jobs actually posted.

thanks! gne1963