• The practitioner is primarily affiliated with a law enforcement body
• There's a profound, firsthand grasp of crime and social disorder data, coupled with the application of the scientific method for data transformations and visual presentations
• Engages in analysis of intricate structured, semi-structured, and amorphous crime data primarily oriented towards criminal justice themes
• Broad utilization of analytical and statistical instruments, including but not limited to IBM’s SPSS, Cognos, Microsoft Power BI, Stata, Statdisk13, and Jupyter Notebook, aimed at algorithm development for regressions and point continuity
• Crafting and publicizing data visual narratives for the judiciary, crime interception, community disorder rectification, and legal/regulatory adherence
• Profound cognizance of data prejudices and the intrinsic constraints present in PredPol
• A pronounced focus on Bayesian and Inferential statistics, aimed at discerning deeper meanings while constantly refining conclusions based on emerging data
• Active engagement in disseminating insights, outcomes, and theories to broaden comprehension of crime and disorder phenomena
• Initiating data metrics bespoke to the agency and embracing methodologies for Exploratory Data Analysis (EDA)
• Curating database frameworks for operational, auxiliary, and administrative tasks

The Relevance of Data Science in Criminal Justice

In our contemporary era, characterized by a strong postmodernist mindset, all knowledge—whether existing, perceived, or anticipated—is persistently scrutinized and often met with skepticism. The foundational certainties and truths, once widely accepted, are now being re-examined. As we navigate the second wave of criminal justice reform, there's a prevailing sentiment that data science and criminal justice are mutually exclusive domains. This perspective, however, is markedly off the mark. While "big data" encompasses information relevant to criminal justice, a significant number of police personnel, from officers to chiefs, remain largely unfamiliar with this burgeoning discipline. Ponder on the term "Data-Driven Policing." What does it genuinely entail? How can law enforcement agencies harness data to shape their strategies? Is it solely about optimizing operational resources? Which specific data sets are taken into account, and what tools facilitate their analysis?

A cursory exploration of online platforms and articles authored by practitioners underscores a conspicuous knowledge gap regarding resources and their deployment within the wider policing community. To put it plainly, many in law enforcement agencies are either oblivious to the importance of data science or, if aware, frequently seek expertise from the private or corporate sectors. This oversight is often rationalized by citing the prerequisites for policing roles, with a common sentiment being a preference for fieldwork over computer-based data crunching. While the priorities of operational, administrative, and academic roles within agencies might differ, the need to comprehend, interpret, and convey insights derived from data is paramount. This emphasis on data-driven strategies is not merely a fleeting trend; its significance will only amplify as we move forward.

How can Agencies use CJDS to Prevent and Solve Crime?

Let’s explore some little-known facts. The 2021 transition to the National Incident Based Reporting System (NIBRS) created opportunities for law enforcement agencies to further standardize incident data collection by gathering Structured Data. These data are used to provide a snapshot of daily, weekly, monthly, and yearly crime by feeding records management programs. Crime analysis apply Frequentist Statistics sifting through Internal Acquisition Systems to inform decisionmakers and afford opportunities for resource deployments to lower the frequency of reported crime. These stratagems are timeless and important when applied correctly. Secondary to crime analysts, intelligence analysts use Data Procurement, or data obtained from outside the organization such as social media posts or financial transactions, to cobble together Unstructured Data to create probability measures to inform decisionmakers through examinations of text for meaning, emotion, and intent. These procedures are highly diverse, reliant on target awareness, and are often limited in design and scope.

In both instances we are looking for things that we already know exist. The total number of carjackings, for example, is structured data we are already aware of. Hate speech by radicalized groups is generally easily identified through pre-defined key phrases, images, and terms which have previously been categorized. What happens when we try to conduct meaning analysis using unknown variables? Where do we turn to when we are looking for something we don’t know we need to look for? This is what separates traditional criminal justice crime analysis from CJDS.

CJDS In Practice – A Hypothetical Example

Why is CJDS so critical? Aside from the points discussed above, law enforcement agencies are beginning to (and will continue to) encounter situations that include big data, for which they are unequipped to manage. Data Silos create object confusion or the intersection of causation and correlation as a fundamental (and totally avoidable) flaw, often resulting in errors. When big data and data silo’s meet, misinterpretation and defective logic are to be expected.

Here is an example of object confusion you can explore within your agency illustrating the need for CJDS recognition. Take arrest rates over time as our example. Submit a request to any crime analysis unit to provide you with the total number of arrests, per day, starting in on January 1, 2020, to January 1, 2023. You should ask for a file containing two categorical variables (columns): date, sum of arrest. Group dates by month.

Create a bar chart with the date on the horizontal axes, and the frequency count of arrests on the vertical axes. Your visual should look something like this:

Do you notice anything of significance? From a frequentist perspective, your results may indicate a marked change in arrest totals starting in March or April of 2020. Typically, crime analysts would seek to explain any change by adding additional variables such as arrest type, patrol area, felony or misdemeanor, adult or juvenile and so on. Even with additional variables an increase or decrease, overall, would still be present. Conclusion, arrest rates and enforcement measures decreased due to events of the time; The Covid-19 pandemic. Object confusion is present.

Expand on the original request by obtaining the same data spanning January 1, 2019, to the current date. Plot these data using the same measures as before.

If Covid-19 caused a decrease in agency arrest rates, why have the overall totals not returned to pre-Covid-19 levels? Is there something else occurring here? What other phenomenon is at play? Criminologists and sociologists would examine social factors to explain change via measures of community stress, unemployment, closure of primary and secondary education centers, state, and national crime rates and so on. Criminal justice data scientists would see the quandary and obtain data from other sources to explain the shift including the same areas criminologists and sociologists would examine. Additional data sources should include agency staffing levels, traffic patterns, residential and commercial construction efforts, response times, fleet management repair backlog, agency application and hiring rates, officer retention benefits, legal changes related to police reform, judicial prosecution rates, court outcomes, agency morale, and budgetary implications (to name a few).

Using CJDS, data scientists understand, have access to, and can obtain structured and unstructured data through internal acquisition systems and / or data procurement through justice partners (such as various courts) to analyze and display potential causes of enforcement change rates.

The field of CJDS is less concerned with data volume as an organizations’ technological ability to collect, store, and retrieve data typically outweighs their aptitude and capability to perform complex and logical situational analysis.

What can an Agency do to Incorporate Criminal Justice Data Science into Operations?

The key take-away from this article is to insert CJDS concept understanding within the culture of law enforcement. All to often academics and practitioners stand on opposite sides of the axiomatic isle both claiming the other is unqualified to significantly contribute to the understanding of criminology. This could not be any further from the truth. CJDS is a massive bridge that stands between practitioner and academic that includes plenty of room for everyone.

Platitudes aside, CJDS is not just a term, or a phrase; it is a procedural mentality, a new focus embedded with complexities and simplicities alike. There are several key components that will put an agency of any size on the right and proper path to CJDS incorporation. Here are a few such steps:

• Education is key. Agency analysts should become well verse in SQL, R, and Python data analysis languages
• Conduct a review of all systems designed to store and capture electronic data and work to create connections between systems with the goal to reduce and / or remove data silos
• Educate the public to dispel big-data rumors. All too often works are published claiming that police agencies are for-profit data storage entities that sell victim data to businesses. Agencies should publicly disclose that they protect collected criminal justice data and that it is never collected for the purpose of profiteering
• Review every element, every variable, every question collected by a records management system to evaluate overlapping concepts, and identify areas often overlooked. As a suggested starting point, I would begin with any data collected during an aggravated assault (NIBRS 13A) incident
• Place emphasis on academic research. This includes contributing to and learning from ongoing works related to how we understand crime and disorder
• Establish connections with other criminal justice data scientists and ask them to review agency software and data collection / analysis systems. Some of us are reclusive (mainly talking about myself, I suppose), but we all share the same deep passion; using data to advance our collective understanding of social problems for the betterment of our communities and the country

Final Thoughts

The specialty of CJDS is, in its own right, worthy of our attention. It intersects three traditional fields within academia and is an emerging area of specialization with untold potential requiring acknowledgement and establishment within our community. Several priorities overlap within the field of criminology, but CJDS is separate and distinct as it places precedence within real-time application based operational and administrative functions to aid in ongoing critical decision processes. Criminology (the study of crime and the criminal justice systems as constructs) is grounded in exploring causality over time; where CJDS supports this exploration and affords immediate operational operability.

Complex criminal justice data demands ethical analysis via Criminal Justice Data Science. Now is the time to embrace and promote systemic best-practice protocols intended for transparent analysis of big data within administrative, operational, and investigative priorities.