Model averaged effect size on [0,1] p and real covariate

[simone77]

Hi,

I have tested the effect of an external covariate, say, flooding surface, and it has been standardized in GEMACO.
I have a set of relevant models that include this effect on my parameter type and I want to model average the effect size of this covariate and finding the effect size with respect to the real (non-standardized) covariate values with its corresponding CI's.
In other words I would like to report the way (magnitude, direction and associated error) my parameter type changes along with the change of one unit of the covariate (on its real scale).

How can I do that starting from the Intercept and slope beta's and their standard error found in the output?

I have been reading several topics here in this forum and both the MARK and E-SURGE manuals (and help on fixed effect) but I am not very confident on how I should proceed to achieve these model averaged estimates.

Thanks for your help,

Simone

[simone77]

I have just realized that even though I am working in E-SURGE, this post perhaps should be moved to the "Analysis & Design questions" (don't believe I can do it). I apologize.

I am looking for mathematical formula to get model averaged estimates of the intercept and slope (with associated error) of an environmental covariate in order to know how much the response variable (say, survival) varies in the [0,1] scale for each unit of change in the covariate in its real scale.
My starting point is: (i) the covariate intercept and beta values in each model in the Logit scale (with their Standard error and CIs) and (ii) the covariate has been standardized.

Thanks for any help,

Simone