A non-technical explanation of omitted variables bias
The economics blog Mahalanobis provides what I think is a nice explanation of the potential bias that results when one excludes relevant variables from a statistical analysis. It is a useful example that, best of all, is explained without jargon or greek symbols. This is more difficult to do than one might think.
posted by david burnstein at 11:45 AM
1 Comments:
It's a nice webpage and I'm going to go back to it. But who would be simpleminded enough to believe that something like quit-rate could be explained, ie, have virtually all non-error of measurement variance accounted for, by one variable?
Post a Comment
<< Home