lmer vs INLA for variance components

November 24, 2018

(This article was first published on R – Insights of a PhD, and kindly contributed to R-bloggers)

Just for fun, I decided to compare the estimates from lmer and INLA for the variance components of an LMM (this isn’t really something that you would ordinarily do – comparing frequentist and bayesian approaches). The codes are below. A couple of plots are drawn, which show the distribution of the hyperparameters (in this case variances) from INLA, which are difficult to get from the frequentist framework (there’s a link to a presentation by Douglas Bates in the code, detailing why you might not want to do it [distribution is not symmetrical], and how you could do it… but it’s a lot of work).

Note that we’re comparing the precision (tau) rather than the variance or SD… SD = 1/sqrt(tau)

As you’d hope, the results come pretty close to each other and the truth:

cbind(truth = c(tau, tau.ri), lmer = 1/c(attr(vc, "sc")^2, unlist(vc)), inla = imod$summary.hyperpar$`0.5quant`)
      truth      lmer      inla
       3.00 2.9552444 2.9556383
group  0.25 0.2883351 0.2919622

Code on Github…

To leave a comment for the author, please follow the link and comment on their blog: R – Insights of a PhD.

R-bloggers.com offers daily e-mail updates about R news and tutorials on topics such as: Data science, Big Data, R jobs, visualization (ggplot2, Boxplots, maps, animation), programming (RStudio, Sweave, LaTeX, SQL, Eclipse, git, hadoop, Web Scraping) statistics (regression, PCA, time series, trading) and more...

If you got this far, why not subscribe for updates from the site? Choose your flavor: e-mail, twitter, RSS, or facebook...

Comments are closed.

Search R-bloggers


Never miss an update!
Subscribe to R-bloggers to receive
e-mails with the latest R posts.
(You will not see this message again.)

Click here to close (This popup will not appear again)