Parameter transformation in sbiofitmixed

Hello,

I have a question regarding sbiofitmixed. I have fitted a model using log parameter transformation. In the NLMEresults object I can see the transformed values of the fixed effects in conjunction with their standard errors. While I can extract the original non-transformed values of the fixed effect by either using exp to transform them back or extracting them from the NLMEResults.PopulationParameterEstimates field, I can not do the same with the standard errors. I could in theory use the Delta method but I have a small sample size so that is not a proper idea here.

My question is if there is a way to extract the back-tranformed Standard Errors of the model fit using sbiofitmixed or the NLMEResults object?

Thank you for your answers in advance.

Best,

Ben

Answer 1

[Florian Augustin]

Hi Ben,

An alternative idea would be to compute a confidence interval for the fixed effects (for example, estimatedValue + [-1.96, 1.96] * standardError). You can then back-transform the confidence interval to get a confidence interval for the population parameters. Would this give you the information you are looking for?

Best,

Florian

[Ben]

Hello,

It solves my problem, thank you! However I am still curious if one can obtain the back transformed values with a function, or is it not possible?

Best,

Bence