The data for 862 societies, at the time before their encounter with Europeans, was published in Ethnology, volume 6, number 2 (April 1967), and then in a hardcover from the University of Pittsburgh Press.
Since then, more societies were added, and some of the 1967 data were revised (see here for detail). The final version, compiled by Patrick Gray in World Cultures, vol. 10, issue 1 (1998), is available as the SPSS data file or R file from this webpage.
- For Stata users, use USESPSS ado (by Sergiy Radyakin) to convert the SPSS data.
The Patrick Gray version contains 1267 societies. However, there are two duplicated observations (Chilcotin and Tokelau).
- According to Gray (1998), which lists the revision history for every society included in the data, one of the duplicated entries for both Chilcotin and Tokelau appeared in the April 1967 issue of Ethnology and had no further revision afterwards. The other entry appeared before the April 1967 issue (October 1964 for Chilcotin; January 1967 for Tokelau).
- Given these, it seems appropriate to keep the entry that appeared in April 1967.
- Whether or not each observation appeared in April 1967 is recorded in the variable v89.
Date of observation
Different societies are observed in different years. Fenske (2013) note that "eight societies [are] observed before 1500 (Ancient Egypt, Aryans, Babylonia, Romans, Icelander, Uzbeg, Khmer and Hebrews)" (page 1366).
The original dataset records the centroid of each society as a pair of integers (longitude and latitude in degrees). To associate each society with geographic characteristics, different researchers use the different definition of each society's geographic `territory'.
- Alesina et al (2013) use a circle of 200km radius around the centroid.
- Mayshar et al (2015) use a circle of 20 mile radius around the centroid.
An alternative approach is to match each society with the map of ethnic groups from another source.
- Michalopoulos and Papaioannou (2013) match with Murdock (1959)'s ethnolinguistic map of Africa, digitized by Nathan Nunn (downloadable from his website)
- Fenske (2013) extend this to the regions outside Africa, by using a variety of ethnic group maps (see section 1.3 for detail).
Location (latitude and longitude), Major subsistence activities (gathering, hunting, fishing, animal husbandry, or agriculture), mode of marriage (dowry etc.), family organization (extended or nuclear, monogamy or polygyny, etc.), type of agriculture (irrigation, type of crops), number of jurisdictional levels, specialization of economic activities by sex, language group, class structure, form and prevalence of slavery, succession rules for local headman, property inheritance rules, type of dwellings.
The data on the number of jurisdictional levels beyond local community is used as the explanatory variable for economic development by Gennaioli and Rainer (2007) and Michalopoulos and Papaioannou (2013). It is also used as the dependent variable by Fenske (2014), Alsan (2015), and Mayshar et al (2015).
Alesina et al (2013) exploit the data on the use of plough in agriculture and on gender-based division of labor in agriculture, to show the correlation between the two.
Fenske (2013) use the variables on the presence of land rights and slavery. Page 1368 lists various other studies using Ethnographic Atlas.
Alsan (2015) use the variables on domesticated animals, female participation in agriculture, intensive agriculture, indigenous slavery as well as jurisdictional hierarchy, to see if these variables are explained by the climate suitability for Tsetse flies (which kill domesticated animals in Africa).
Boix (2015, chapter 1) looks at the correlation of foraging societies with the size of settlements, inheritance rules, social stratification, and the number of jurisdictional levels beyond local community.