I´m quite familiar with the use of PGLS in R (I´m using Caper package) but I don´t know how to implement confidence intervals. I´ve searching for information in the web and it´s quite difficult. I read in this paper that are people using some kind of CAIC package which I cannot really found: "Human frontal lobes are not relatively large" in PNAS 2013.
Well, my question is if someone could give me some guideline in how to create the confidence intervals of a PGLS in R. Thanks!
https://github.com/JeroenSmaers/evomap
this new package calculates CI and PI for a PGLS.
Well, you can easily get confidence intervals for model parameters. Do you mean fitting a confidence band around the regression line? This is not possible in general as the intervals depend on the phylogenetic position of the species to be predicted. You could construct confidence bands assuming the species are independent, but that is why you are not using OLS regression in the first place. See
http://biology.ucr.edu/people/faculty/Garland/GarlIv00.pdf
I have no idea about the evomap package.
Well, the idea should be to create the "confidence band" around the regression line. I saw this here: http://www.pnas.org/content/110/22/9001.full.pdf?with-ds=yes but I can´t find the package and tutorial.
I can't find the package either. I suspect it is an early version of caper. However the confidence bands look really strange in that paper: they are always parallel. One usually finds that the bands are wider in the extremes of the data. The authors don't say how they arrived at these results. I am always amazed at how users can seem to defy the laws of mathematics. See fortunes::fortune(203) and fortunes::fortune(211)
- Badges
- Science topic