SCCS Precursors

From InterSciWiki
Jump to: navigation, search

Immediate precursors of Murdock and White (1969)

Precursors of the SCCS (Murdock and White 1969, Standard Cross-Cultural Sample (see wiki page Standard Cross-Cultural Sample) are White's SRAS:-- Societal Research Archives System: Retrieval, Quality Control and Analysis of Comparative Data. (Social Science Information 7(3): 78-94). SRAS was the first compendium of machine-readable coded data, see 42 SRAS datafiles. It included Murdock's Ethnographic Atlas of 1270 societies, from which the sample of best-described societies in the earliest time periods was drawn, and the 200 World Sampling Provinces (Murdock 1968 Ethnology, Vol. 7, No. 3 (Jul., 1968), pp. 305-326), designed as sampling strata to separate geographic and cultural clusters with much greater sharing and mutual influence within rather than between them.

In the SCCS, some provinces were again combined by this criterion. The SRAS contained a matrix analysis of the overlap between independently drawn samples used by different cross-cultures and showed that the pairwise deletion of cases required for combining variables from different authors was so great that no cumulative cross-cultural research was possible based on combined samples.

The purpose of the SCCS was to build a sufficiently large sample -- each society pinpointed in a time and space where the ethnographic or historical data was of the best quality -- to enable multivariate analysis and to provide the ethnographic bibliography necessary for future researchers to contribute cumulative data (codes on different variables for societies in the sample) so that analysis could combine the work of different contributors. For the SCCS, Murdock and White (1969) divided the work in preparing pinpointing sheets and ethnographic bibliography ranked from primary to auxiliary and secondary sources.

On-line update: Murdock, George Peter, and Douglas R. White. 1969 [new edition 2002]. Standard Cross-Cultural Sample. Ethnology 9: 329–369. http://eclectic.ss.uci.edu/~drwhite/pub/SCCS1969.pdf

The Tylorian "Method of Adhesions"

As Murdock (1968:305) noted, "Modern cross-cultural research stems from two innovations in the comparative ethnographi methods made by Tylor (1889): (1) substitution of the comparison of "adhesions" or intra-cultural associations for the comparison of isolated cases wrenched from their cultural contexts and (2) the use of statistics to test the bearing of the discovered associations on scientific hypotheses." This was the principle of Murdock's (1949) classic work on Social Structure (New York: Macmillan Press). He attempted to selecties societies for his sample that were as diverse as possible to avoid the criticisms leveled at Tylor's method of adhesions.

Critiques of Tylor's method of adhesions: Nonindependence

As statistician Sir Francis Galton had shown in his oral critique of Tylor's (1889) paper, however, the "modern" principle of adhesions was a not sufficient basis on which to draw scientific inferences when the sample of cases included regional clusters of societies with close similarities, or clusters of similar societies in the same language family. i.e., copies of the same cultural prototype. The effect of sampling nonindependent cases was to deflate the number of "effectively independent" cases and inflate significance tests of correlations so that the scientific inference that these were non-random might be drastically incorrect. This is know scientifically, subsequent to Galton, as the problem of autocorrelation.

Corrections for nonindependence and attention to autocorrelation in Murdock and White (1969)

Naroll (1961, 1964, Naroll and D'Andrade 1963)and others went on to show the effects of Galton's problem and autocorrelation and suggest tests and corrections for errors of statistical inference due to autocorrelation. White (1969) had shown autocorrelation effects in his study of North American "trait adhesions" and added one of Naroll's tests of autocorrelation to Murdock and White (1969) as the foundational article for a successor to the "modern" approach to statistical inference. Tests of independence and nonindependence for the four SCCS variables which they included in their article showed that the "effectively independent sample sizes", as a function of the taking every n'th societies in the 186 society SCCS to get statistically independent cases (i.e., lack of autocorrelation effects) were:

N=62 Economy (Sampling interval 3)
N=46 Political Integration (Sampling interval 4)
N=21 Language (Sampling interval 9)
N=19 Descent (Sampling interval 10)
Map of Indigenous Language families of 186 societies in the SCCS (Standard Cross-Cultural Sample)
A map of the Indigenous Language families of societies in the SCCS for example, shows a distribution of numbers of societies of the same family to vary from very large to small clusters or isolated, and thus a large potential for autocorrelation from societies of the same language family.

References to Solutions to Galton’s Problem prior to Murdock and White (1969)

Naroll, Raoul. 1961. Two Solutions to Galton’s Problem. Philosophy of Science 28:15-39.

Naroll, Raoul and Roy G. D'Andrade. 1963. Two Further Solutions to Galton’s Problem. American Anthropologist 65: 1053-1067. http://www.jstor.org/stable/668583

Naroll, Raoul. 1964. A Fifth Solution to Galton’s Problem. American Anthropologist 66:863-867. http://www.jstor.org/stable/668182

See: jstor articles

White, Douglas R. 1969. Cooperation and Decision Making among North American Indians. Ann Arbor, MI: Dissertation Reprints.

Incorrect attributions

Sources citing the Murdock and White Standard Cross-Cultural Sample "from the Human Relations Area Files (HRAF)" (Berry, Poortinga, Segall, and Pandey 1997:173) are factually incorrect. The sample was chosen independently of any other sample, HRAF included, and the codings done by Murdock and White were from original sources, not the HRAF files. Some of the bibliography overlaps with sources in the HRAF, but that is all. Murdock had founded HRAF but disavowed its policies long before the SCCS project was begun. There was a short period circa 1986 where HRAF distributed the SCCS datafile but HRAF was not the creator of SCCS.

Berry, John W. Ype H. Poortinga, Marshall H. Segall, Janak Pandey. 1997. Handbook of Cross-cultural Psychology. Vol 2: Social behavior and applications. Boston: Allyn and Bacon.