Policing and Mental ill-health: Using Big Data to Assess the Scale and Severity of, and the Frontline Resources Committed to, mental ill-health-related calls-for-service

Abstract

Addressing public safety and welfare, inclusive of responding to incidents involving persons with mental ill-health (PMIH) has become an integral dimension of, and a significant challenge to, contemporary policing. Yet, little is known of the scale and severity of such PMIH-related policing demand, nor of the extent of frontline resource consumed in resolving such incidents. To address this shortfall, we deploy a bespoke text mining algorithm on police incident logs to estimate the proportion and severity of calls-for-service involving PMIH in a study of Greater Manchester, UK. Furthermore, and using Global Positioning System data, we then assess the amount of time spent by frontline officers responding to these calls. Findings suggest that existing police recording practices serve to significantly underestimate the scale and severity of PMIH-related demand. The amount of time spent dealing with PMIH-related incidents is both substantial and disproportionate relative to other forms of police demand.

Shedding Light on the Dark Figure of Police Mental Health Calls for Service

“…In other words, there are over ten times as many calls for service in our study jurisdiction that involve PwPMI than if one were to look at the 'Mental Health' call classification alone. This finding is in line with those of Langton et al (2021) who found that 9.9% of calls for service in Manchester, England involved PwPMI even though the 'Mental Health' call classification only comprised 1.9% of all calls for service. Together, these findings show that police call classification practices based on primary nature do indeed mask a significant proportion of calls for service that involve PwPMI.…”

“…We find that while the proportion of calls for service in our data under the 'Mental Health' call classification (n = 397, 0.9%) is consistent with the ~1% rate identified in previous research (Hodgkinson & Andresen, 2019;Koziarski, 2020;Langton et al, 2021;Livingston, 2016;Lum et al, 2021;White & Goldberg, 2018), our results reveal that PwPMI were, in fact, present in 10.8% (n = 4,646) of all calls in 2019. In other words, there are over ten times as many calls for service in our study jurisdiction that involve PwPMI than if one were to look at the 'Mental Health' call classification alone.…”

“…For instance, Langton and colleagues (2021) examined 3.6 million calls for service made to the Greater Manchester Police in England between January 2014 and September 2017 and found that PwPMI calls-identified through a specific call classification-only comprised 1.9% of all calls during this time, which is largely in-line with previous research (Hodgkinson & Andresen, 2019;Koziarski, 2020;Langton et al, 2021;Lum et al, 2021;White & Goldberg, 2018). However, by using an artificial intelligence approach to mine calls for service text logs, ignoring flags or addresses associated with any particular call, they instead found that a striking 9.9% of calls for service involved PwPMI (Langton et al, 2021). In other words, their analysis revealed that PwPMI are involved in approximately five times more calls for service than are identified through the 'Mental Health' call classification alone.…”

“…Recent estimates typically suggest that approximately 1% of all calls for police service involve PwPMI (Hodgkinson & Andresen, 2019;Koziarski, 2020;Langton et al, 2021;Livingston, 2016;Lum et al, 2021;White & Goldberg, 2018).…”

The spatial (in)stability of mental health calls for police service

“…Second, calls for service data are commonly classified on the basis of what the primary issue of a call was (Hodgkinson & Andresen, 2019b; Langton et al., 2021; Shore & Lavoie, 2018; Vaughan et al., 2016). This data‐management practice consequently masks secondary, potentially important components of each call.…”