Why Use a Biased Estimator?

[thrillhouse86]

Hey All,

I am comfortable with the idea of biased and unbiased estimators, but what I don't understand is why you would ever want to use a biased estimator ? at the end of the day doesn't it mean that the sample statistic is different from the population statistic you are trying to estimate ?

 

[D H]

but what I don't understand is why you would ever want to use a biased estimator ?

The simple answer is that the biased estimator might be all you have available. Lots of sensors yield biased measurements. The sensor model had better include an estimation of the bias to yield an answer closer to the truth than that produced directly by the sensor.

 

[thrillhouse86]

So does that mean you never really want a biased estimator, but sometimes practical issues force you to work with one ?

 

[D H]

Exactly.

 

[thrillhouse86]

cool thanks D H