Does Consistency Matter? An Integration of Neighbourhood Modelling and Suicide Clustering
Item statusRestricted Access
Spatial suicide research largely informs where local action plans and preventative strategies to reduce the number of suicides should be targeted. After reviewing the literature, it was found that contemporary studies on the spatial patterning of suicide largely rely on using only one administrative geographic unit in their analysis. This means that there is currently no method in place to assess the validity or robustness of the results generated. One important consideration which is largely overlooked directly attributable to this decision is that results stemming from this analysis will be inherently sensitive to the modifiable areal unit problem (MAUP) which means that these studies may have been incorrectly guiding local suicide prevention policies. Using Glasgow, Scotland as a case-study, this research paper presents findings which suggest that the use of only one administrative areal unit in cluster analysis does have the ability to under-report suicide incidences. This was investigated further by introducing automated zoning techniques as a mechanism for modelling neighbourhoods at different scales and configurations as a way to account for variance attributable to the MAUP. It was found that the neighbourhood models out-performed the administrative units in terms of the frequency of clusters identified and cluster consistency with respect to location. The mechanism behind this was largely attributable to population uniformity within the synthetic geographies which was not a feature of Scottish administrative geographies. No relative benefit was discovered by introducing socio-economic variables as a proxy for modelling neighbourhoods in terms of the cluster analysis results. Additionally, this paper presents a discussion as to the underlying weaknesses with current clustering techniques.