Crime and Social Media Sentiment in Jamaica: A Predictive ARIMAX Time-Series Analysis (2010-2025)

Abstract

This study examines the predictive relationship between social media sentiment, macroeconomic indicators, and weekly youth crime in Jamaica using an Autoregressive Integrated Moving Average with Exogenous Variables (ARIMAX) framework for the period 2010-2025. Official crime data were obtained from the Jamaica Constabulary Force (JCF), while social media sentiment was derived from geolocated Twitter and Facebook posts using the VADER sentiment analysis algorithm. Macroeconomic covariates included gross domestic product (GDP), unemployment rates, and fuel prices obtained from national statistical authorities. Stationarity was assessed using the Augmented Dickey-Fuller test, and model selection was guided by Akaike and Bayesian Information Criteria. The optimal ARIMAX (1,1,1) model demonstrated strong explanatory power, explaining approximately 70% of the variance in weekly youth crime. Results indicate that negative social media sentiment significantly predicts increases in youth crime, while GDP growth and fuel prices show positive associations, and unemployment exhibits a negative relationship. Temporal persistence was confirmed through significant autoregressive and moving average components. Findings suggest that integrating digital sentiment analytics with macroeconomic monitoring enhances predictive crime modelling and supports early warning systems for youth violence. Policymakers should consider incorporating real-time social media analytics into national security intelligence frameworks.