Interpretations of prehistoric social organization based on multivariate statistical analyses of burial practices are becoming increasingly common in the North American archaeological literature. Unfortunately, these analyses and interpretations can incorporate weaknesses ranging from faulty data coding and the mis-application of statistical procedures to biases in the statistical and logical procedures employed. These problems are discussed in light of recent analyses (Tainter 1975a, 1975b, 1977a, 1977b, 1978) which use burial data from six Woodland sites in the riverine midwestern United States to develop a model of social change for the period A.D. 200-800. The results of these particular analyses are shown to be, at best, highly ambiguous and, at worst, contradictory to the proposed interpretations. This paper summarizes the weaknesses in these analyses, both to show the absence of support for the particular proposed interpretations and to illustrate how inappropriate methods can negate potentially useful mortuary research.