If you are referring to what is typically presented in Monte Carlo simulations, then the relative error is calculated simply as: simulated mean - true value. The absolute relative error is abs(simulated mean - true value).

If you mean something different, please provide more detail, and perhaps an example.

Ariel

I would say that relative error is absolute error divided by the true value. This is a reasonable measure in many cases, because a large true value doesn't usually need such precise estimates as a small value. But there are times when a large relative error is acceptable--especially if a variable could have either sign. When estimating a probability near zero, relative error is not particularly useful.

Let me add a couple more measures to my earlier post:

relative percent error: (simulated mean - true value)/true value.

absolute percent error: abs(simulated mean - true value)/true value

In addition to simulated values, you can use these formulae for any predicted vs actual values. You just to know the actual and predicted, or the true vs estimated...

I hope this helps

Ariel

to give a simple concrete medical example:

Suppose blood pressure is measured to the nearest 1 mmHg due to quality of instrument and human eye, then . absolute error = 1mm Hg.

for a blood pressure reading of 200 mmHg relative error is 1/200 = 0.5% or 0.005;

for a blood pressure reading of 100 mmHg relative error is 1/100 = 15 or 0.01.