Mutiple Sources in the Collection of Data on Political Conflict

This paper examines the costs and benefits of multiple source coverage in the context of measuring political conflict since 1960 within 30 black African countries. Results indicate that if we follow the common approach to this problem and make comparisons across data sets that are based on different sources or combinations of sources, then apparent source coverage effects are observed. However, it is difficult to attribute these effects uniquely to variations in coverage. In fact, when we disaggregate our own data on collective protest by source, we observe no pronounced source coverage effects. Similarly, no such effects are found in a model of the determinants of coups d'etat. This suggests that the cost-benefit ratio of multiple sources may be unnecessarily high, and that the measurement implications of previously made distinctions between global vs. regional sources and visible vs. less visible forms of political conflict may have been overstated.