Hostname: page-component-cd9895bd7-hc48f Total loading time: 0 Render date: 2024-12-27T10:59:36.115Z Has data issue: false hasContentIssue false
  • English
  • Français

Statistical Classification1

Published online by Cambridge University Press:  01 July 1940

A. L. Kroeber
A. L. Kroeber*
Affiliation:
University of California, Berkeley, California
Get access Rights & Permissions [Opens in a new window]

Extract

I present herewith four-cell coefficients calculated from data given in two recent midwestern archaeological reports by Webb and Griffin. The purpose is not controversial, since in a simple situation involving a small number of culturally related sites a competent archaeologist saturated in his material can draw all important classificatory inferences and it would be vain to hope that any statistical device would add anything fundamental. What calculations can do in such a case is this: (1) Check errors and oversights of “intuitive” or inspectional interpretation. This of course holds both ways: if there is serious disagreement, a coraputatory error may be at fault. (2) Present results with added clarity and incisiveness, especially after numerical coefficients have been translated into diagrams. (3) Indicate minor revisions of classification. (4) Sometimes suggest the factor at work if a classification comes out conflicting or dubious at certain points.

Type
Research Article
Information
American Antiquity , Volume 6 , Issue 1 , July 1940 , pp. 29 - 44
Copyright
Copyright © The Society for American Archaeology 1940

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

Footnotes

1

Typewriting, computations, and diagram by a University of California W.P.A. project, No. 665-08-3-30, Unit A-15.

References

1 Typewriting, computations, and diagram by a University of California W.P.A. project, No. 665-08-3-30, Unit A-15.

2 Webb, W. S., An Archaeological Survey of the Norris Basin in Eastern Tennessee. Bureau of American Ethnology, Bulletin 118, 1938.—Griffin, J. B., An Analysis of the Fort Ancient Culture, Notes from the Ceramic Repository … of the University of Michigan, No. 1, 1935.

3 As my colleague and collaborator, H. E. Driver, has shown, all formulae give more or less misrepresentative results when one or two of the four values a, b, c, d differ very greatly from the others, that is, when their distribution is markedly skew. According to the nature of the particular skewness encountered, each coefficient can be specially misrepresentative, speaking broadly.

4 Actually only seventy three show variation: two are universal positives (rectangular post-mold patterns, clay floors) and should not have been counted. I noted them as universal only after the computations had been made; but they are too few to affect the coefficient relations materially.

5 Two of these, 4–6 and 5–17, seem to have been erroneously counted, either by Webb or by me.

6 Webb says, page 372, that this list includes traits occurring at two or more of the six sites; all traits occurring on only one site having been eliminated “as having no value in this comparison.” The last phrase may be doubted. He is not trying to prove similarity but to establish degree of similarity, and on this degree, unlikenesses, even if unique, certainly have a bearing.

7 They might, of course, be prehistoric Cherokee, from a time when Cherokee culture had not yet attained its historic form.

8 It is of course possible that the Thomas sites are really similar to the Webb sites in pottery, to Harrington and Nacoochee in the remainder of the culture; but it would be hazardous to assume this without intensive reexamination of all the material recovered.

An addendum has been issued for this article:

Addendum