[R-sig-ME] variance explained by fixed effects in MCMCglmm

[rafter sass ferguson]

I've been searching for ways to calculate some R^2-like statistics for a
multi-level multi-response model fit with MCMCglmm (with 3 Gaussian
responses).

I can see from the Course Notes and elsewhere that it's very
straightforward to calculate the variance explained by the random effects,
but after much searching I haven't found a discussion for fixed effects. It
looks like one option would be refitting with all my fixed effects as
random - but I'm concerned that might be bonkers, or that there might be an
easier way.

For previous multilevel modeling with lmer, I've used the approach
following Nakagawa et al. 2013 (A general and simple method for obtaining
R2 from generalized linear mixed-effects models. Methods in Ecology and
Evolution, 4(2), 133–142. http://doi.org/10.1111/j.2041-210x.2012.00261.x)
to calculate conditional and marginal R^2...
but I'm not sure how to adapt it for the (for me) brave new world of
MCMCglmm.

I would be grateful for any advice!

Thanks so much for your time.

Warmly,
Rafter


Rafter Sass Ferguson, MS
PhD Candidate | Crop Sciences Department
University of Illinois in Urbana-Champaign
liberationecology.org
518 567 7407

[[alternative HTML version deleted]]

[Krause, R.W. (Robert)]

Hey Rafter,

I had a similar question some time back and came across this webside.

http://jonlefcheck.net/2013/03/13/r2-for-linear-mixed-effects-models/

Here functions for calculating marginal and conditional R² are provided.
My knowledge about these mathematics and philosophy is limited so it might be that someone on this list has good arguments not to use these functions or calculate these R²s at all.
If so, please step forward.

Best regards,
Robert

Robert Krause, MS
Behavioural Science Institute
Radboud University Nijmegen


________________________________________
Von: R-sig-mixed-models [r-sig-mixed-models-***@r-project.org]" im Auftrag von "rafter sass ferguson [***@gmail.com]
Gesendet: Donnerstag, 16. Juli 2015 01:05
An: r-sig-mixed-models
Betreff: [R-sig-ME] variance explained by fixed effects in MCMCglmm

I've been searching for ways to calculate some R^2-like statistics for a
multi-level multi-response model fit with MCMCglmm (with 3 Gaussian
responses).

I can see from the Course Notes and elsewhere that it's very
straightforward to calculate the variance explained by the random effects,
but after much searching I haven't found a discussion for fixed effects. It
looks like one option would be refitting with all my fixed effects as
random - but I'm concerned that might be bonkers, or that there might be an
easier way.

For previous multilevel modeling with lmer, I've used the approach
following Nakagawa et al. 2013 (A general and simple method for obtaining
R2 from generalized linear mixed-effects models. Methods in Ecology and
Evolution, 4(2), 133–142. http://doi.org/10.1111/j.2041-210x.2012.00261.x)
to calculate conditional and marginal R^2...
but I'm not sure how to adapt it for the (for me) brave new world of
MCMCglmm.

I would be grateful for any advice!

Thanks so much for your time.

Warmly,
Rafter


Rafter Sass Ferguson, MS
PhD Candidate | Crop Sciences Department
University of Illinois in Urbana-Champaign
liberationecology.org
518 567 7407

[[alternative HTML version deleted]]

_______________________________________________
R-sig-mixed-***@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models