Measuring city relationship strength beyond total counts: A multidimensional framework for distinguishing prominence from interdependence and significance

Relying on counting the number of interactions to gauge city relationship strength can be misleading, as volumes often only reflect the prominence of large cities rather than city interdependence. Drawing on statistical concepts of effect size and confidence, this study develops a relationship classification framework that identifies interdependent and statistically significant relationships. For demonstration, this framework is applied to placename co-occurrences in English Wikipedia articles for 100 European cities. Each city relationship is evaluated through five metrics: co-occurrence, mutual information, statistical confidence, a combined mutual information–confidence metric and a relative gravity model. The findings demonstrate that a high co-occurrence, commonly observed between large cities like London and Paris, typically corresponds with high statistical confidence, but does not necessarily imply strong interdependence. By contrast, strongly interdependent relationships tend to be regionally clustered, such as the Dutch Randstad (Amsterdam–Rotterdam–The Hague), the Flemish Diamond (Brussels–Antwerp–Gent) and the Ruhr region (Dusseldorf–Essen–Duisburg). By differentiating relationship types, this framework reveals the complexity of intercity relationships and regional patterns that conventional methods fail to capture, offering a more nuanced understanding of city networks.