I know only one definition of bias in statistics: the expected difference between the estimator and the parameter. Are there any alternative definitions?


Omega Contributor
Correct, bias is the difference between the estimated and true value of a parameter. In study design and analyses there are lots of other types of bias (e.g., measurement error, missing data, etc.), but the statistical definition is as mentioned. Just picture the bull's-eye example, where the estimate is either missing the target or not.


Ambassador to the humans
Hlsmith be careful with your wording there. It isn't the difference between the estimated value and the truth. If that was the case then any "unbiased" estimator would always get it exactly right. It's the difference between the expected value of the estimator and the true value.


Omega Contributor
Thanks. I almost wrote it that way, but got lazy. Since I had thought about Best Linear Unbiased Estimators and also how least absolute shrinkage and selection operator is not an unbiased estimator.