Many years ago, as part of my MSc in GIS, I undertook a dissertation exploring the relationship between crime, poverty, and domestic violence.
Domestic violence (DV) has been a growing issue in Northern Ireland, with increasing rates recorded since the centralisation of records in 2005. It is defined as: “Any incident of controlling, coercive, threatening behaviour, violence or abuse between those aged 16 or over who are, or have been, intimate partners or family members regardless of gender or sexuality.”
This study employed Geographic Information Systems (GIS) alongside statistical techniques—specifically, Ordinary Least Squares (OLS) Regression and Geographically Weighted Regression (GWR)—to explore relationships between poverty, various crime types, and domestic violence. The results confirmed the well-established link between poverty and DV, but also revealed significant associations with other crime categories.
Unlike most studies, which focus on selected regions, I chose to analyse all of Northern Ireland at the Electoral Ward level. This decision stemmed from the lack of previous country-wide DV research. Northern Ireland, being relatively small in size and population and featuring both rural and urban settings, seemed an ideal case. In hindsight, this scale proved overly ambitious due to the extensive data processing involved. A more focused study on urban centres, such as cities, might have been more manageable.
The core aim of the project was to determine whether DV in Northern Ireland is statistically linked to poverty and other crime types. The use of GIS and statistical modelling helped to uncover spatial and temporal patterns, and to identify the most influential socio-economic and crime-related predictors. While DV is underreported globally and locally, prior research has seldom examined it at national scale or in relation to broader crime trends—particularly within the UK. This study helped address that gap.
Methodology
- Data Sources: PSNI DV reports (2005–2017), crime statistics, deprivation indices (NIMDM), and census data.
- Geographic Units: Wards (Ward1993 and Ward2014 formats); areal interpolation was used to convert data between formats.
- Analytical Tools: ArcGIS (v10.5.1) for spatial analysis, SPSS (v25) for statistical modeling.
- Statistical Methods:
- Bivariate correlation (Pearson’s)
- Ordinary Least Squares (OLS) regression
- Geographically Weighted Regression (GWR)
Key Findings
Strongest DV predictors:
- Criminal damage, violence/sex/robbery, and social housing were consistently strong predictors.
- Low vehicle ownership showed a high correlation with DV.
- Temporal patterns: DV rates have risen steadily since 2009, with 397 of 462 wards reporting increases.
- Spatial patterns: Higher DV rates were clustered in urban areas such as Belfast and Derry-Londonderry.
- Model performance:
- OLS models showed high adjusted R² values, indicating strong predictive capability.
- GWR provided better local model fit but was sensitive to data quality and geographic unit issues.
- Data issues: Converting between old and new ward boundaries introduced potential inaccuracies due to the Modifiable Areal Unit Problem (MAUP).
Limitations
- DV data limited to annual reports with no detailed time, demographic, or geographic breakdown.
- Underreporting of DV likely.
- Conversion between spatial datasets may have introduced errors.
- Lack of individual-level data limited granularity.
Conclusions
There is a significant correlation between DV, poverty, and other crime types.
Socio-economic indicators (like lack of education, vehicle ownership, and social housing) are strong predictors. This research supports more targeted resource allocation to reduce DV in high-risk areas.