Accuracy, stability and reciprocity in informal conversational networks in rural Kenya

Abstract

Abstract In this paper, we ask several questions about conversational networks. How accurately do respondents report the characteristics of their network partners? How stable are these networks over time? And if one participant in a conversation reports it, does the other? The data come from a longitudinal survey of ego-centered conversational networks and reproductive behavior in small villages in rural Kenya. Because network partners were also interviewed in one of the four research sites, we are able to compare the respondents’ reports of their alters’ behavior with these alters’ self reports. We find a strong relationship between the reported behavior of egos and their reports of their alters’ behavior, as others have done elsewhere. However, discrepancies between the respondents’ reports of their network partners contraceptive behavior and their network partners’ self reports are substantial. Levels of network stability and reciprocity were also low. The discrepancies between the respondents’ reports of their network partners and the partners’ own reports are primarily due to incorrect reporting by the respondents, whereas the low levels of stability and reciprocity are related to the difficulties of measuring regular but diffuse social structures. Some of our results are specific to the context of rural Kenya, but other are more likely to be generalizable. The influence of social networks on behavior may be strong or weak, but its measurement has systematic problems.