Open GitTFJ opened 3 years ago
I don't think the /
syntax is supported in phyr
(am I right on that @daijiang ?), but (1|time/species)
is equivalent to (1|time) + (1|time:species)
. An interaction in the grouping term is also not supported by phyr
I don't think (?), but you could easily construct a factor that is equivalent to time:species
and use that as your grouping term, e.g (1|time) + (1|time_species)
, where time_species
has a unique value for each combination of time and species. Likewise this could be done for each nesting level that you need, e.g. (1|time) + (1|time_species) + (1|time_species_site) + (1|time_species_site_region)
. In each case the variable has a single unique factor value for each combination of the grouping terms being nested. Here is a good blog post that shows three ways to code nested random effects (where the method I suggest, which should definitely work with phyr
, is the third way: "implicit nesting"): https://www.muscardinus.be/2017/07/lme4-random-effects/
Hope that helps.
Is there a way to include multi-level nested random effects e.g. (in lme4 syntax) (1|time/species/site/region)?