Open MoralesCastilla opened 2 years ago
After conversation with Will Pearse and Jonathan, it seems like our initial interpretation of lambda and sigma was not off.
Lambda is analogous to Pagel's but instead of modelling phylogenetic structure in model residuals at once, we sort of deconstruct it and get an estimate of phylogenetic structuring for the responses to each cue. (I guess in other settings, analyzing different response variables where the major driving factors for that response and where only some of those are included as model predictors, lambda values could be biases. However, in our case, it should be similar to partitioning a PGLS lambda).
Sigma measures the variance of the response to each cue accummulated through time (~rate of evolution). Potential biases could arise as with Lambda, but we are not running specific tests for this.
From meeting today, we suggested we do not need to do any additional analyses to show lambda is analogous to Pagel's, but we should perhaps include a short discussion of this somewhere and link to simulation scripts.
Check notes in PMM repo and ask @lizzieinvancouver and @legault if in doubt to make sure we come with a satisfactory interpretation for sigmas (and lambdas) in the model.