Measuring Changes in Corruption Over Time

How confident should we be in the answers we get when using existing measures of corruption? Scholars routinely rely on country-level corruption indices to study democratic accountability, representation, development, and institutional reform. Yet a great deal of research argues that country-level corruption indicators are incomparable to each other and over time. We present evidence that these indicators all track a shared component identifiable as a change within country in overall corruption, but despite this, they remain problematically inconsistent with each other. As a result, we argue that empirical findings about corruption based on these designs will be particularly sensitive to the choice of measure. To address this problem, we introduce a new synthetic measure of within-country change in corruption that is robust to bias and measurement error that may be present in those existing measures. Our new measure allows researchers to explain changes in corruption that are common to all measures, rather than privileging any single index. We employ our new measure to revisit a published analysis of the effect of women’s representation on corruption, finding a more consistent relationship and more balanced dynamics using our measure than with others’.