7
$\begingroup$

What is the difference between identifiable and estimable?

I know that when two models are identifiable they must have distinct parameters in the corresponding parameter space. But what is the formal definition of estim-ability? Say, if given a one-way ANOVA, $\mu-\tau_{i}$ are all estimable, so are their linear combinations. But how about $\frac{\mu-\tau_{i}}{\mu-\tau_{j}}$? This is not an estimable parameter since it is not a linear combination of those $\mu-\tau_{i}$'s. But it is still identifiable, right?

I want to know what is the difference and the relationship between these two notions.

$\endgroup$
0

1 Answer 1

2
$\begingroup$

Identifiability is related to the mathematical model without consideration for any real-world noise in the observations.

Estimability takes measurement noise into account.

Jacquez, J.A. & P. Greif (1985) "Numerical Parameter Identifiability and Estimability: Integrating Identifiability, Estimability, and Optimal Sampling Design", Mathematical Biosciences 77(1):201-227.

$\endgroup$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.