[R-sig-ME] Is it ok to use lmer() for an ordered categorical (5 levels) response variable?

Peter Claussen d@kot@judo @end|ng |rom m@c@com
Thu Mar 7 03:57:05 CET 2019

> In the backyard: fallow area, nettles (*Urtica dioica*), ivy (*Hedera helix*),
> and brambles (*Rubus spp.*) are each scored one if present, and the
> naturalness index was computed as the sum of these scores.
> => it results in a 5-levels ordinal variable because it can go from 0 to 4,
> and each increase in 1 means a backyard with more features of 'naturalness'.
> I wonder thus if this could be modelled using a glmer() with family =
> binomial and feeding to the model two columns: cbind(sum of 1's, sum of
> 0's) (see R documentation for family{stats}, in the Details: "*As a
> two-column integer matrix: the first column gives the number of successes
> and the second the number of failures.*")
> I will try and see how the model fit the data. But I would be interested in
> getting a theoretical opinion.

I’m not sure summing the scores is correct. Are the distributions of nettles, ivy and brambles independent, or could they be correlated? 

If you have a fallow area and find nettles, is it also likely to have ivy? If so, the the binary outcomes (nettles present and ivy present) are not really additive.  Similarly, are the outcomes equivalent. That is, is a fallow plot with nettles, and nettles only, equal in ’naturalness’ to a fallow plot with ivy and only ivy?

You might want a multivariate approach. Have you considered principal components to reduce each set of (nettles, ivy, branbles) to a single score, and then analyze that score using lmer?



	[[alternative HTML version deleted]]

More information about the R-sig-mixed-models mailing list