Analysing outcome variables with floor effects due to censoring