1

Which is a more serious problem in regression, regression dilution or bias?

Regression dilution usually causes underestimation, and bias (especially selection bias)usually causes overestimation. Which problem is more severe?

In practice, which case is more often, underestimation or overestimation?

user2986288
  • 51

1 Answers1

3

In practice it doesn't make sense to ask for general patterns in whether dilution or "bias" [1] is more important. Some problems are subject to neither, some are subject to only one, and some are subject to both.

[1] What kind of bias? Regression dilution is a kind of bias, so how can we compare regression dilution and "bias" in general?

Paul
  • 10,920
  • Do you agree that "Regression dilution usually causes underestimation, and bias (especially selection bias)usually causes overestimation. "? – user2986288 Jul 18 '15 at 14:28