repolr: An R Package for Fitting Proportional-odds Models to Repeated Ordinal Scores

Modelling repeated ordinal score data is a common statistical problem, across many application areas. The proportional-odds model is widely applied to such repeated ordinal scores and can be fitted in the repolr package (repeated measures proportional odds logistic regression) in R using the method of generalized estimating equations (GEE). The GEE approach specifies a model for the mean of the correlated observations within clusters of repeated scores for each individual without fully specifying the joint distribution of the observations. This paper describes the core features of package repolr, which has undergone extensive changes since the first release in 2008, for the first time. A number of example datasets and extensive R code are used to illustrate a range of data analysis tasks that users of repolr may typically wish to undertake.