Pseudo-Rsquareds

Compute the Nagelkerke and Cox--Snell pseudo-rsquared statistics, primarily for logistic regression. A generic function with methods for glm and svyglm. The method for svyglm objects uses the design-based estimators described by Lumley (2017)

psrsq(object, method = c("Cox-Snell", "Nagelkerke"), ...)

Arguments

object

A regression model (glm or svyglm)

method

Which statistic to compute

...

For future expansion

Value

Numeric value

References

Lumley T (2017) "Pseudo-R2 statistics under complex sampling" Australian and New Zealand Journal of Statistics DOI: 10.1111/anzs.12187 (preprint: https://arxiv.org/abs/1701.07745)

See also

AIC.svyglm

Examples

data(api)
dclus2<-svydesign(id=~dnum+snum, weights=~pw, data=apiclus2)

model1<-svyglm(I(sch.wide=="Yes")~ell+meals+mobility+as.numeric(stype), 
     design=dclus2, family=quasibinomial())

psrsq(model1, type="Nagelkerke")
#> [1] 0.2397372