When we hear or read about crime trends, we tend to take such news and statements at face value. Politicians, journalists, and pundits alike are quick to cite figures that lack any context. In Mexico, monthly crime figures are paraded across the media, tweeted and retweeted, and often cited to support certain position or policy.
One recurring example is the unfortunate and adverse practice of relying on extortion and kidnapping figures. These crimes have dark figures (i.e. unreported crime) above 98%, meaning less than 2% of crimes are reported, much less investigated and prosecuted. Any change in the number of reported cases for those crimes could mean that more (or less) people are coming out to report a crime, and not necessarily that more (or less) crimes are being committed. In fact, a rise in reported extortion cases could actually be a good thing, as it could mean that more people are willing to call the police and report a crime. This is rarely discussed or at least mentioned in the mainstream media or its pundits, and should be seriously taken into account when informing the wider public, as well as in policymaking.
Another common mistake is using absolute number of cases to describe crime trends. Saying that the State of Mexico had the greatest number of X crime wouldn’t be that misleading if it weren’t for the fact that the state is home to about 14% of the country’s population. Unfortunately, it’s still uncommon to account for population differences when comparing crime trends; to this day some people still talk-tweet about crime using absolute numbers instead of rates.
Crime rates, however, have their own downsides. Since rates are a measure of risk – in that it represents the probability that an individual, a thing (e.g. a car) or location (e.g. a business) can become a victim of an offense – these depend wholly on using the proper numerator (e.g. the count for some crime) and denominator (the at-risk population).
For starters, it’s crucial to specify which type of offense is being measured when selecting a numerator. For example, a rate for auto theft is confusing if it includes all types of vehicles or if it doesn’t specify which types of vehicles are being considered. Does the rate include only cars, or also trucks, buses, motorcycles, and other types of vehicles? If data is available only as an aggregate of all vehicle thefts, then the rate should be for vehicle theft in general.
Furthermore, it’s important to use the proper denominator to describe the different risks of victimization in a population. The typical rates used to track crime trends assume that all individuals have the same probabilities of victimization. For most offenses, however, that is clearly not the case. People of certain demographics and socioeconomic levels run higher risks of becoming victims of certain crimes. Individuals’ routine activities also matter; if you take the subway from your home to work and pickpocketing is common in your route, you and your fellow riders have an additional risk of victimization. If women are more often reported as victims of pickpocketing in that same route, then that’s an additional level of risk.
One more example: a rate for auto theft of 250 thefts per 1,000 inhabitants is uninformative at best. Since not all people drive cars, it’s misleading to use the overall population as the denominator for calculating the rate. A better way to do so would be using the number of cars in circulation, which is most often available as the number of insured cars on the road. Although still a flawed denominator for overall car theft (e.g. it doesn’t account for uninsured cars, which are plentiful in Mexico, for example), this rate offers a more accurate snapshot of auto theft.
An upcoming post will delve into an alternative measure for analyzing crime: the Location Quotient (LQ).