Fluoride and dental caries: Two different statistical approaches to the same data source

A recent analysis of data from earlier papers on the relationship between dental caries and drinking water fluoride concentration suggested that the commonly accepted inverse relationship did not exist. Our reanalysis of those data, however, confirms the well‐known association between fluoride concentration and dental caries. It also shows that the contrary result arose misleadingly from three simultaneous methodological errors: use of a unifactorial instead of a multifactorial model; omission of or over‐aggregation of some data, and analysis of homoscedastic probits instead of heteroscedastic counts.