Mislabelled default LR test

[pitakakariki]

> simdata$g <- as.character(simdata$g)
> fm <- lmer(y ~ g + (1|x), data=simdata)
> doTest(fm, fixed("g"))
boundary (singular) fit: see ?isSingular
p-value for predictor 'g': 1.434264e-13
          --------------------
Test: Kenward Roger (package pbkrtest)
> doTest(fm, fixed("g", "lr"))
boundary (singular) fit: see ?isSingular
p-value for predictor 'g': 1.434264e-13
          --------------------
Test: Likelihood ratio
> doTest(fm, fixed("g", "kr"))
boundary (singular) fit: see ?isSingular
p-value for predictor 'g': 1.035445e-10
          --------------------
Test: Kenward Roger (package pbkrtest)

[pitakakariki]

Error seems to be in defaultdesc, which checks for factor but not character.

[pitakakariki]

Might apply to lm too - t-test vs LR.