SFI Human Galaxy

From InterSciWiki
Jump to: navigation, search


The Human Galaxy featured at various research centers worldwide provides educational software to access the five major databases of coded ethnographic (and textual) data, including the Murdock-White SCCS (Standard Cross-Cultural Sample), Lewis R. Binford LRB/WFS (World Forager Sample), Murdock EA (Ethnographic Atlas), and Jorgensen WNAI (Western North American Indians) plus Anthon Eff's XC database that combines variables from all these four into one having societies with variables from two or more others. Peter Turchin and his colleagues are coding a new time-series database for historical and archeological empires, in which many of the communities represented in the previous five datasets are spatially and temporally located. Michael Fischer, VP of eHRAF, is working to include ethnographic texts alongside the societies with ethnographic codes in these six databases, and HRAF's president Carol Ember is cross-indexing database codes and textual categories.

The Santa Fe Institute posts links at option to insert this Online SFI Human Galaxy link at SFI to provide access to statistical analysis of coded data from SCCS, EA, LRB/WFS, WNAI and XC. Nearby SFI in Santa Fe, the School for Advanced Research in Santa Fe allows researchers to also apply for access to the SAR library option to insert this Online Human Galaxy link at SAR that provides access to ethnographic sources and additional access to a 72 Society SCCS book collection and other SAR books pertaining to possibly 30 or more additional temporally pinpointed SCCS societies. Researchers may apply for these permissions to dearden[at]sarsf.org or LHolt[at]sarsf.org. Researchers who have their own university's subscription to eHRAF or are affiliated with the Santa Fe Institute may request to use the SFI eHRAF subscription and will find the online and library resources complementary. The SAR collection (and potentially, other libraries), together with eHRAF, may provide the textual materials for new SCCS coding projects for single or groups of variables that can be submitted to the journals such as Cross-Cultural Research (Sage), World Cultures or Structure and Dynamics (eScholarship). The latter two are open access (Creative Commons CC-BY-NC Noncommercial) for authors, or other journals.


An example of cross-cultural research on the study and map of high god world religions, the 2014 PNAS Carlos_A._Botero#EA (Ethnographic_Atlas) Map - PNAS - PDF of the text - estimates the relative effects of resource abundance, ecological risk, cultural diffusion, shared ancestry, and political complexity on the global distribution of beliefs in (moralizing) high gods. The EA codebook variable v34 for High Gods "Supportive of human morality" (value 4) versus others (Absent, Not active in human affairs, Active in human affairs but not supportive of human morality) shows 181 High Gods out of 748 coded by Murdock (1967). Codebooks for other datasets show that of 186 societies coded in the Standard Cross-Cultural dataset (SCCS variables v2001, v2002), 19 are deep Islam and 7 superficial Islamization, while 6 are deep Christian and 24 superficial Christianization), while fewer than 10 comprise other examples of high god religions. Extrapolating to the PNAS map, 83% of the high god societies are likely to be Christian or Islamic and perhaps 10% of others whose high gods were the result of prior Christian/Islamic proselytizing unrecognized by the ethnographer. One may conclude from these two congruent samples (the SCCS being a subset of the EA) that most of the PNAS findings for high gods apply to stronger or weaker forms of Christianity or Islam. The Brown and Eff 2010 model for high gods (SCCS) is at http://capone.mtsu.edu/eaeff/DEf_SCCS.html.
Screen Shot 2014-11-11 at 6.47.14 AM.png

As in most cross-cultural research, the high god world religions study illustrates the nonindependence of human societies. Fortunately, for the first time since 1889, this problem, known as Galton's (networks) problem, was recently solved by Dow-Eff functions abbreviated as DEf (e.g., Brown and Eff 2010 and Eff and Dow 2009), and is cited by Botero (2014: ref. 9), who uses analogous procedures to control for autocorrelation. The Dow-Eff solution is implemented in the R language and made tractable for researchers and courses in the Human Galaxy. A two-stage regression model creates a single variable, Wy, which when added to a second stage ordinary regression may control for autocorrelation if it succeeds in creating random error terms. This, along with other tests, is tested statistically. DEf can be implemented in R but is easier to use in a classroom friendly CoSSci Galaxy for the SCCS, EA, WNAI, LRB and XC datasets. It is an open access site at http://SocSciCompute.ss.uci.edu/ where students or researchers can enter variables from these datasets (see codebooks at http://capone.mtsu.edu/eaeff/DEf_SCCS.html) to estimate models with imputation of missing data and correction for autocorrelation.



Contact Douglas R. White for further information