Measurement and Data Aggregation in Small-n Social Scientific Research

How should small-n researchers aggregate the information collected during their research in an effort to measure the relevant theoretical concepts with high levels of validity and reliability? This article specifically focuses on the method of triangulation, which is frequently used in process-tracing approaches. We introduce and theorise different aggregation strategies commonly used in triangulation, such as weighted and simple averages or ‘the winner takes it all’ strategy. We then evaluate their performance with regard to their proneness to measurement error using computer simulations. Our simulation results show that averaging different information sources, in general, outperforms other aggregation strategies. However, this is not the case if poorly informed sources are biased in a similar direction; in these situations the ‘winner takes it all’ strategy shows a superior performance.