Theory of Change
Connecting cities through standardized data can promote diversion policies
Our theory of change leverages Code for America's skills in crowdsourcing, custom-built software, and data analysis to identify opportunities to lower police involvement in certain non-violent 911 calls. Nation-wide visibility into 911 call volumes and the ability to identify peer cities (using varied criteria) allows us to promote diversion opportunities for the most significantly impacted communities.
We begin by addressing the fundamental standardization barrier. Once 911 data is standardized across cities it is cross referenced against census and demographic data.
Reimagine 911 and industry partners work together to identify patterns in 911 call volumes, the most significantly impacted individuals, suitability of call types for alternative response, and peer cities.
Reimagine 911 works with partners and advocates to identify diversion policies that have succeeded locally, and to find peer cities that could adopt similar policies. Advocates are supplied with data and case studies and pursue system change locally.
1. Connecting a Fragmented 911 System
A theme we heard from 911 domain experts was that the hyper-local nature of 911 makes it difficult to understand 911 across jurisdictions. Since there are over 6,000 individual 911 PSAPs (public service answering points), coding similar incidents differently across each jurisdiction can be an insurmountable barrier when researching opportunities for system change.
The Reimagine 911 program believes it has identified a method to overcome this hurdle.
Reimagine 911 has demonstrated a viable process for aggregating call-for-service records and standardizing call types in order to associate similar incidents across cities. In the network diagram below we can see four cities whose local call types have been associated with standard APCO codes and are now interconnected.
Reimagine 911 has completed the preliminary step of standardizing local call types. Although we have not yet undertaken the next step: connecting our standardized call-for-service dataset to census and demographic data, we have laid the technical groundwork for that cross-referencing.
The Classifyr software we use for crowdsourcing standardized call types is built on top of an ETL pipeline architecture. This pipeline allows us to add new data transformations that add, alter, or filter the dataset that is ultimately generated. While some technical work remains to create a transformation to retrieve and associate demographic data with each 911 call, the larger framework for doing so is in place.
2. Multi-City Analysis
Once a rich dataset has been created, Reimagine 911 will perform data analysis with the help of a research partner who works in the emergency response sector. We have been tracking some Potential Research Questions in our conversations. Some of these questions relate directly to this Theory of Change regarding diverting 911 calls away from police. Many others do not though, which demonstrates the potential for knowledgable domain experts to introduce valuable paths of inquiry that we have not yet considered.
Some immediate analysis on our initial dataset is possible. Although the underlying data needs to be vetted and cleaned further to support any conclusive observations, analysis on this early data does illustrate what might be possible.
The interactive visualization below shows the prevalence of standardized calls for each of 5 cities. These standardized APCO codes generally include 1 to 8 local call types that are unique to each city.
Try This: click the APCO code "NEGLECT" to isolate that standardized call type. Next select "New Orleans" and other cities to compare the prevalence of local 911 calls related to neglect in those cities.
Previously it would have been quite laborious to identify all the related local call types needed to compare relative call volumes across different cities. A standardized dataset allows multi-cony comparison with much greater ease.
After incorporating demographic data it should be possible to identify relative volumes of 911 calls that represent the types of human impact and benefit to the public we are most interested in supporting:
Calls such as
MENTAL HEALTHthat may be diverted to existing non-police services.Calls like
TRESPASSINGthat disproportionately affect vulnerable populations.Variance along social indicators such as poverty and education.
Systemic trends such as declines in 911 call volume following high profile police-related deaths.
3. Promoting Diversion Policies
In the same way that 911 call standardization allows us to associate call volumes across cities, it can also promote sharing diversion policies across cities.
Example: In July of 2022 the 988 Mental Crisis Hotline was launched as a non-police resource in all 50 states. Proponents across the country are trying to establish policies that allow 911 call takers to divert incoming calls to 988. With a standardized call dataset, advocates could look at cities with successful policies and identify "peer" cities using different criteria (population, historical call volumes, etc) most relevant to the cities, the callers, and the call type.
Reimagine 911 could work with a variety of partners and advocates eager to promote policies from from other cities. Reimagine 911 could supply them with data and case studies that would help make the case for system change locally.
Last updated