Question

I once saw the GLMM modeling building process using the following script:

dative.glmm8 <- lmer(RealizationOfRecipient ~ AnimacyOfRec + DefinOfRec +
PronomOfRec * PronomOfTheme + I(AccessOfRec=="given") + AnimacyOfTheme + DefinOfTheme +
I(AccessOfTheme=="given") + log(RatioOfLengthsThemeOverRecipient) + (1|Verb),      
family="binomial")

I do not understand the passed argument of “I(AccessOfTheme==”given”)”? What is the physical meaning of this kind of argument setting?

Answer 1

This question is not actually lmer-specific, but applies to all model formulas in R. In a formula context, I() stands for “insulate”: from http://cran.r-project.org/doc/manuals/R-intro.pdf ,

I(M ) Insulate M. Inside M all operators have their normal arithmetic
meaning, and
that term appears in the model matrix.

This is essentially creating a dummy (0/1) variable on the fly for AccessOfRec being equal to “given” (1) or anything else (0).

You could also do this by creating the variable beforehand, e.g. AccessOfRec_given <- (AccessOfRec=="given"), and then using the derived variable in the formula.

By the way, I would strongly recommend using the data argument to lmer, rather than either using variables from the global workspace or attach()ing data frames.