Within this framework, two different interpretations of the residual class are possible that can lead to different numeric values for the same data. The first is that the residual class is composed of numerals that need to be removed to obtain results consistent with the presumed distribution of accurate results. The second is that the residual class is composed of numerals that would need to be changed in order to obtain a distribution consistent with the assumed distribution of accurate results. The first interpretation leads to a special case of the Rudas et al. (1994) “mixture index of fit.” The second interpretation leads to a measure related to the Gini (1914) dissimilarity index.