Difference between real estimates and derived parameters

[KristaHupman]

Hi all,

I am new to MARK and was wondering if there is someone who can explain what the difference is between the real estimates and derived parameters that can be exported from MARK to excel?

Many thanks!

[cooch]

Hi all,

I am new to MARK and was wondering if there is someone who can explain what the difference is between the real estimates and derived parameters that can be exported from MARK to excel?

Many thanks!

In very short form, 'beta estimates are the estimates from the linear model, which are typically (by default) calculated on either the sin or logit scale. The are commonly referred to as 'beta' estimates, since they are the estimated 'slopes' (beta terms) of the linear model. The 'real' estimates are 'reconstituted' (back-transformed' from the transformed scale to the 'real' probability scale, which is generally the 'scale' of interest for most purposes.

Derived parameters are parameters which are calculated as a function of other parameters, where those 'other' parameters are included in the likelihood. For example, if you fit a Pradel 'survival and recruitment' model, then survival (phi) and recruitment (f) are in the likelihood (and thus can be constrained to be linear functions of things of interest), whereas lambda is 'estimated' by algebra (as lambda = phi + f). As such, lambda is a 'derived' parameter. The derived parameters are reported with SE, and can be exported to Excel along with any other sort of parameter (generally, in MARK, *everything* can be exported in one form or the other -- betas, reals, derived parameters etc.).