Abstract: |
Interest in the material correlates of economic specialization has led to numerous quantitative studies of standardization (or the lack of it) in craft production, particularly of ceramics. Most of this research has focused on measures of variation and conventional statistical procedures in the treatment of the empirical data, many of which are dependent on unrealistic assumptions (such as normal populations), yielding results that can be questioned. To resolve this crisis more robust statistical methods are investigated including the "jackknife" method (for confidence interval construction) and a multigroup test for homogeneity of variance known as the Brown-Forsythe Test. Computer simulations show that the latter is robust under a variety of distributional forms. These new methods are used to reanalyze ethnoarchaeological ceramic data from the Philippines
it is shown that markedly different conclusions can be reached when compared with the results of more conventional procedures.
|