We develop an original dataset designed to characterize traditional gender institutions for pre-colonial societies in Sub-Saharan Africa. The data extend several variables derived from the Standard Cross Cultural Sample (Murdock and White 1969) to a more geographically-dense set of ethnic groups in the African continent included in Murdock (1967) Ethnographic Atlas. The resulting dataset captures traditional gender data for 322 ethnic groups. We collect our final dataset through (1) literature-based and (2) survey-based data collection.
Our primary source of ethnographic data is derived from the sociology and cultural anthropology literatures. We use a set 412 Sub-Saharan ethnic groups drawn from Murdock (1967)’s Ethnographic Atlas as our sample. We extend the bibliography of ethnographic references for the Ethnographic Atlas with any additional literary sources with the potential to cover traditional ethnographic characteristics for this set of societies. Our preferred ethnographic coding process follows the methods outlined by Whyte (1978). We form a gender-balanced1 team of 12 undergraduate research assistants. In an effort to mitigate bias associated with engendered interpretations of the primary literature, each research assistant is paired with another team member of the opposite gender. The sample is divided equally across each research assistant pair who, for each ethnic group assigned, are tasked with independently reading through the references listed in our bibliography and encoding a set of cultural characteristics on gender characteristics into standardized categorical variables2. The research assistants in each pair then reconvene to resolve their independent codes into a single encoding. Where they are in agreement, that code is the final literature-based code. Where they disagree, they try to come to an agreement based on the information available. Where the pair could not agree, the code is set to “Ambiguous”.
A principal investigator and a lead research assistant (one female, one male) piloted the preliminary coding procedure and monitored the codings generated by the undergraduate research assistant team. Any codes needing review were identified and discussed with the undergraduate pair responsible. A final systematic review of all the literature-based codes was conducted to identify errors and maintain consistency3.
Of the 412 Sub-Saharan ethnic groups in our sample derived from Murdock (1967), 307 groups had some form of literary sources with information pertinent to the present study. To gather data for the remaining groups, we develop an electronic survey to mimic the variables coded in the literature-based phase. The remaining uncoded groups, along with a random 10% sub-sample of the groups already coded by our undergraduate research assistants as a quality check, constitute a “survey-based” sample.
We identify potential survey respondents as individuals who can credibly comment on pre-colonial characteristics, including academic researchers, members of the ethnic group itself, and other cultural experts. Potential respondents were approached through an outreach campaign to complete the electronic survey of their own volition. Survey responses were then cleaned for consistency with the literature-based data.
One meaningful difference in the survey-based phase is that unlike the literature-based encoding, we could not feasible train each potential survey respondent on each definition to the same level as we did for the undergraduate research assistant team tasked with literature-based coding. For example, we were able to explain and ensure that our preferred definition of a solidarity group was completely understood among the undergraduate team4. We address this by including definitions and examples for each variable considered by the survey so that each respondent can better understand each specific question.
Murdock, George Peter. 1967. “Ethnographic Atlas: A Summary.” Ethnology 6 (2). JSTOR:109–236.
Murdock, George Peter, and Douglas White. 1969. “Standard Cross-Cultural Sample.” Ethnology 8 (4). JSTOR:329–69.
Whyte, Martin King. 1978. “Cross-Cultural Codes Dealing with the Relative Status of Women.” Ethnology 17 (2). JSTOR:211–37.
6 females and 6 males↩
This set of variables can be viewed in the spatial visualization of the dataset and the Supplemental Information page.↩
Details on this process can be found in Supplemental Information page.↩
In addition to an initial research assistant training session, explanations and examples were often discussed in detail in person and via email exchanges.↩