Spatial Modeling of Graffiti as a Function of Street Network Centrality: A Case Study in San Francisco

Graffiti presents serious urban concerns, often signaling urban decay. This study uses open spatial data to analyze and model graffiti occurrences in terms of street network centrality measures. In particular, betweenness centrality, closeness centrality, and degree centrality are evaluated using San Francisco, California, as the case study area, with data from OpenStreetMap and reported graffiti from 2008 to 2023 from the San Francisco nonemergency municipal service (denoted as 311) as the data sets. The spatial error model was found to outperform both ordinary least squares tests and the spatial lag model. The model could further explain graffiti spatiality. Graffiti writers were observed to favor street segments that are close to the downtown and well-connected to other streets, often having high accessibility, visibility, and accommodating street furniture. The results indicate that bridges and highway segments that are difficult to stop and tag were typically avoided. In addition, for a given street, the model error in adjacent streets significantly (<i>p</i> &lt; 0.001) affected the observed value.