# Should measurement errors and uncertainty be accounted for in every statistical test? [closed]

Most measurements have measurement errors.

1. Should measurement errors and uncertainty be accounted for in every statistical test?
2. Are duplicates better than single measurements if the sample size is large enough?
-
1. Ideally, yes. This is usually the interpretation of the error term in models. 2. If you are interested in intra-class (or within-individual variation) then duplicates are better than single measurements. –  Macro Aug 1 '11 at 15:18