Hi Jaime, Yes, (p)gls regressions can handle categorical and/or continuous predictors in the same way as conventional regressions do. The (p)gls regressions differ from conventional one (ols) by relaxing the hypothesis of independance (and homoscedasticity) of the residuals.
Regarding which one is better, this depends on the phylogenetic or autocorrelation structure in the residuals of the model fit. Sometimes the use of a gls regression based on a wrong model for our data (e.g. the default Brownian motion option in (p)gls) can be worse than using ols regression. Typically, this can't be known before hand. In practical terms, I think that using models more flexible than brownian motion (BM) - i.e. models that can jointly estimate/accommodate the structure in the residuals of the regression such as OU or Pagel's lambda/mixed models - are usually better because they can handle scenarios ranging from no autocorrelation (ols) to various levels of phylogenetic signal (including BM). In this sense, the use of (p)gls regressions can be considered as generally better. There's also implementations in "phylolm" package, or "ape/nlme" that also allow the use of various models for the residuals structure. Best wishes, Julien ________________________________________ De : R-sig-phylo <r-sig-phylo-boun...@r-project.org> de la part de JAIME ADRIAN CORDOVA <jaime.cord...@wisc.edu> Envoyé : mercredi 19 juin 2024 20:46 À : r-sig-phylo@r-project.org <r-sig-phylo@r-project.org> Objet : [R-sig-phylo] Predictor types for caper:pgls Hi all, I'm using the caper package to run some pgls regressions, with the goal of understanding if the species being studied has any relation to the growth rates (e.g. Does being an E. Coli strain mean it grows better in our environments of interest than if it was Salmonella?) Can a categorical predictor be included in the pgls function of caper? Whether it would be on its own (continuous ~ categorical) or with a continuous predictor (continuous ~ continuous + categorical). I recognize that the brunch function is available but I would like to do this using pgls. From what I understand, pgls is better? Thanks! Kind regards, Jaime Jaime A. Cordova Jr. Pronouns: he/him/his/�l Genetics PhD Candidate, Perna Lab University of Wisconsin, Madison [[alternative HTML version deleted]] _______________________________________________ R-sig-phylo mailing list - R-sig-phylo@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-phylo Searchable archive at http://www.mail-archive.com/r-sig-phylo@r-project.org/
