The online Oxford Language Dictionary defines confirmation bias as "the tendency to interpret new evidence as confirmation of one's existing beliefs or theories." Since it is impossible to eliminate one's existing beliefs or theories, the only workaround is to expand the ability to see things in other ways.
I explore some self-explanatory variables for crime pattern recognition in the book Elements of Crime Patterns. Anyone working at the ground level will see that these variables are how we might begin to recognize a serial crime in the data. They are common sense and I documented them because they have not been documented broadly as existing variables to identify. I used these variables daily to look for crime patterns, and many analysts do that. Many of these variables will mean nothing to a researcher, so when reading the book, they could dismiss them. Some researchers might find variables in this section to explore for new research. For example, is there a correlation between masked robbers and the level of violence used in crimes? Many questions could be developed and explored. Whether they matter is another story.
The officer collecting information at the street level and the investigator following up with more information-gathering have needs that conflict with researchers and policy-makers. They want to get their jobs done and may focus more on finding a charge that will stick or getting to a higher-priority call or case. In documenting information, each individual will decide what they believe is essential, which will vary. The quality of data collection suffers because of this.
From The Criminal Investigation Process, Volume III, Observations and Analysis (click on link), October 1975, page 152:
Things may have improved since 1975, but with over 18,000 law enforcement agencies in the United States and little training on the importance of uncovering crime patterns, including the understanding of elements of modus operandi, it is likely that much information falls to the wayside and is lost forever.
So, where does confirmation bias apply here? You are unlikely to find what you do not know much about or value.
Researchers and policy-makers will see the need to understand aggregate problems to find lasting solutions and distribute resources in the best places. They have different objectives than those on the ground.
The citizen who does not want their car broken into will value the agency that uncovers a specific car break-in pattern involving a drug addict stealing from many vehicles in their neighborhood before their car is broken into. They want the police to study a seemingly urgent matter that a researcher or policy-maker is less concerned with. But if the researcher or policy-maker is that citizen, they will want the same.
Those working in law enforcement agencies will have biases based on their expertise. The statistician sees things in terms of statistics, the Excel expert sees the application of spreadsheets, the GIS expert sees geography and maps, and the investigative or intelligence analyst sees research on specific criminals and their actions.
How we focus on crime patterns will depend on where we stand in the world. But once we know our own viewpoint, we can try on others' views and appreciate their value.
Having more than one person or department analyze data can provide different perspectives that may help offset individual biases. Assembling a team with diverse backgrounds and perspectives can be an effective countermeasure against confirmation bias.
Additional reading: Knowledge management, criminal investigation and the Basic Policing Model: an integrated approach towards effective detection (click on link)
No comments:
Post a Comment