Notes on the bias of dissimilarity indices for incomplete data sets: the case of archaelogical classification

Abstract

The problem of missing data is particularly present in archaeological research where, because of the fragmentariness of the finds, only a part of the characteristics of the whole object can be observed. The performance of various dissimilarity indices differently weighting missing values is studied on archaeological data via a simulation. An alternative solution consisting in randomly substituting missing values with character sets is also examined. Gower's dissimilarity coefficient seems to be the least biased one either with 25% missing values and 49%; it has not however a constant behaviour as to the sign of the bias. The simulation experiment has also shown that when average linkage cluster analysis is performed on an incomplete data set either using Gower's index or randomly substituting missing values gives satisfactory results while the modified indices fail to detect the cluster structure