I am running some Monte Carlos where I am interested in observing the
bias in parameter estimates across manipulated conditions. By bias I
mean the absolute percentage difference of the simulated value from the
true value.
I was wondering whether there has been another written about how much
bias is "acceptable"--I know that this is like asking how long is a
piece of string and that there is no statistical fiat that can give a
definitive answer, because it also is a very field specific issue. I
guess in physics or social sciences we would be willing to put up with
different degrees of bias.
What I am looking for what is an commonly acceptable limit in social
sciences (and econometrics) in particular. Does anyone know of any
references in this regard? I am looking at setting the boundary of
"acceptable bias" at about 20%--would that be too high? I welcome any
thoughts.
