Modelling the Impact of Media Coverage on Violent Crime: An Autoregressive Distributed Lag [ARDL] Time-Series Analysis Using a Media Crime Coverage Index

Abstract

The media influence public perception and potentially shape behavioural responses to crime. This study examines whether media intensity contributes to crime escalation using a dynamic time-series econometric framework combined with weighted variable importance analysis. A Media Crime Coverage Index (MCCI) was constructed using keyword frequency analysis and normalised by total media output to measure relative crime-related attention. Monthly data for robbery, assault, and homicide were analysed over ten years using autoregressive models, long-run multipliers, Granger causality tests, structural break analysis, and standardised weight decomposition to assess relative predictorsand their importance. The empirical models control for unemployment and police expenditure to isolate the independent effect of media exposure. The results indicate that lagged media intensity has a positive and statistically significant effect on robbery rates and weaker positive associations with assault and homicide. The weight decomposition shows that MCCI carries the highest relative contribution in the robbery model (weight ? 0.42), exceeding unemployment and police expenditure, whereas unemployment dominates the homicide model (weight ? 0.37). Police expenditure consistently exhibits negative weights across models, confirming its deterrent role. Long-run multiplier estimates demonstrate that media effects are amplified over time due to crime persistence. Granger causality confirms temporal precedence with bidirectional feedback. Structural breaks reveal stronger media impacts during crisis periods. Model diagnostics confirm robustness and statistical stability. The findings suggest that media systems operate as dynamic participants in crime ecosystems with measurable relative importance across crime categories.