0

If I have a sample of rats in which I want to measure protein levels in several different brain regions, the rats are of two genotypes and both sexes, and I want to take into account that the brain regions come from specific animals, which is the appropriate syntax for modeling this with lme4?

marker ~ genotype * sex * brain.region + (1|animal/brain.region)
marker ~ genotype * sex * brain.region + (animal|brain.region)
Bryan
  • 741
  • 1
  • 4
  • 15
  • 3
    A random intercept would be `marker ~ genotype * sex * brain.region + (1|animal)`. You can then add random slopes (correlated or uncorrelated) as appropriate for your data, e.g., `marker ~ genotype * sex * brain.region + (brain.region|animal)`. See [Table 2 of the package vignette](https://cran.rstudio.org/web/packages/lme4/vignettes/lmer.pdf) for an explanation of the syntax. See also http://stats.stackexchange.com/questions/13166/rs-lmer-cheat-sheet – Roland Aug 04 '15 at 16:03
  • Do you have multiple measurements within each brain region of each rat? – Michael M Aug 04 '15 at 16:51
  • Right now, I have no measurements, just rat brain regions and the foresight to plan my experiment and statistics before doing a single measurement. I'd like to assay multiple subsamples per region. If that is not possible, then time and budget constraints might only allow one per region. – Bryan Aug 04 '15 at 18:20
  • @Bryan: See http://meta.stats.stackexchange.com/q/832/17230. Moderators can't promote other users' comments into answers. But start by telling Roland that you think his comment's good enough to be an answer! – Scortchi - Reinstate Monica Mar 18 '16 at 14:53

0 Answers0