This list below is written in terms of what is deemed to be most important to least important changes.
Response-specific random intercepts can now be included in a model. If you are familiar with including random row effects in a model, then one may think of response-specific random intercepts as similar to this, except the row effects are now response-specific specific rather than a common across all responses e.g., species-specific random intercepts to accounted for sampling design on a per-species basis. This is done via the
ranef.ids argument in the main function
boral with downstream uses elsewhere also. Please see the help file for
boral, as well as the new
about.ranefs help file for more information.
Quite a bit of editing has been done to the help files, as well as some very minor names changes to returned objects. For help files, note that the mathematical notation for latent variables has been changed from
u_i, while species-specifc random intercept "covariates" are now represented by
coefsplot function now implements plotting for fourth-corner coefficients, if a fourth corner model is fitted and
fourthcorner = TRUE is set. Please see the help file for
coefsplot for more information.
A new argument
formula.X is now available in the main
boral function. Like other regression functions such as
glm, the argument
formula.X should be an object of class "formula", which represents a symbolic description of the covariate matrix one wishes to include. Please see the help file for
boral for more details.
ALL functions to calculate information criteria and log-likelihood values of any sort are now classed as defuncted and no longer maintained...and probably do not work properly!