Measuring generalised trust in sub-Saharan Africa : a critical note

Abstract

"Generally speaking, would you say most people can be trusted or that one must be careful in dealing with others?" For the past 50 years this question has been used extensively and almost exclusively as a measure of generalised trust in both national and cross-national studies. However, it was not until very recently that scholars focused on the question's validity and reliability as a measure of generalised trust. Besides that these studies' findings are largely contradictory, few of them examine the validity and reliability of the trust data in the African context. This study is motivated by this research gap and the fact that the levels of trust from the Afrobarometer surveys seem to challenge what the literature suggests about the causes and consequences of trust. The study finds that the question is a reliable measure of trust in 'most people' since it obtains largely similar country level estimates when used alone over a period of time. However, African respondents do not consistently interpret 'most people' as 'non-co-ethnics' as previous studies have suggested. In addition, the question does not alternate very well with other measures of bridging trust. This measure is also weakly correlated with measures of civic engagement and associational membership than its alternative, the trust in non-co-ethnics question. However, both measures produce expected linkages with measures of ethnic diversity, economic development and democracy.