Crime Beyond Reports: Estimating the Hidden Crime Landscape with Data Science (WG-PLACE)

• In Event: WG-PLACE 2: Research on crime, place and time

Abstract

A substantial share of criminal activity goes unreported, distorting crime patterns and limiting the effectiveness of policy responses. Traditional data sources - such as police records and victimization surveys - capture only incidents that are officially documented or voluntarily disclosed, often overlooking the broader spectrum of unreported crimes. This study introduces a data-driven framework for estimating hidden crime by integrating multiple data sources to bridge the gap between reported and actual crime levels.

Using advanced statistical modelling and machine learning, this study examines how socio-economic conditions, trust in law enforcement, and local demographic factors shape crime visibility. By comparing recorded crime patterns with indicators of community-level underreporting, the analysis reveals systematic biases that create significant gaps in conventional crime statistics. Findings identify areas where low reporting rates or inconsistencies in data collection may obscure the true prevalence of crime, particularly in marginalized or underserved communities.

This framework enhances crime measurement by providing a more accurate and equitable representation of crime patterns. It also provides actionable insights for policymakers and law enforcement agencies, enabling better resource allocation, targeted interventions, and strengthened community engagement. By bridging the gap between documented and actual crime levels, this research advances criminological theory while reinforcing the need for data-driven strategies to enhance justice and public safety.

Authors

• Jon Bannister, Manchester Metropolitan University

• Monsuru Adepeju, Manchester Metropolitan University, M15 6BH, United Kingdom

• Karolina Krzemieniewska-nandwani, Manchester Metropolitan University, M15 6BH, United Kingdom