16 to a dealer’s 10: Could blackjack odds help inform police pursuit policies?
Data can provide the path through seemingly tough police decisions
By Jonathan Aronie and Geoff Alpert
In 2015, the New Orleans Police Department (NOPD) implemented a highly debated policy restricting the authority of police officers to engage in high-speed vehicle pursuits for non-violent crimes. The debate about the policy pitted community members focused on reducing collisions and injuries to innocent civilians (and to the officers themselves) against some officers who thought it was unconscionable to allow a fleeing suspect to get away.  The debate continues in cities across the U.S., often drawing emotional charges of ignorance and naïveté from one side against the other.
What’s particularly interesting about the police pursuit debate is how often it is presented as a choice between (a) something that is good and (b) something that is bad. At a recent policing conference, an officer passionately argued in favor of NOT restricting police pursuits because, he stated, letting a bad guy go was dangerous. In his mind, if letting bad guys go is bad (and most agree it is), then pursuing them must be good. Those on the other side of the debate adopt an equally “good or bad” perspective. If high-speed pursuits put so many innocent people (including police officers) at risk, which is bad, then restricting high-speed pursuits (in the context of non-violent offenses at least) must be good.
But the good/bad construct, which so often is used to frame this particular debate, is misleading. Like so many decisions, the pursuit/no pursuit debate is not between good and bad. It’s between bad and worse. Is letting criminals go bad? Yes. Is pursuing non-violent criminals at high speeds in urban areas bad? Yes. They both are bad. Framing the question as one between good and bad interferes with our ability to engage in constructive debate in pursuit (if you will) of a sensible answer.
Reframing the pursuit/no pursuit debate
The better way to think about the vehicle pursuit debate is to think about it as a bad hand at a blackjack table. Let’s say you are sitting in your favorite Vegas casino and have just been dealt a 16 (the worst hand possible). You moan aloud, thinking your plight could not worsen until you look up and see the dealer has a jack showing (the best card possible for her.) Your plight just got worse. You are very likely going to lose. If you hit (take another card), you likely are going to bust. (Of the 52 cards in a deck, 28 of them (54%) will put you over 21, causing you to lose.) Alternatively, if you don’t take a card, then the dealer likely will end up with a higher hand than you, and you’ll lose anyway. (If you don’t play blackjack, we offer a short explanation in the footnote below of why this is such a bad situation. )
Remember, the dealer in our example is sitting pretty. She has a 10 in front of her for all the world to see. The most likely card sitting under that 10 also happens to be a 10 because all face cards count as, you guessed it, 10. (Thus, there are 16 “10 cards” in a deck versus 4 of each other card.) And even if she doesn’t draw a 10, there is no card she can draw that will put her over 21.
So, do you hit, or do you stand? If you hit, you likely are going to bust. If you don’t hit, you likely are going to lose to the dealer’s better hand. What a dilemma. Your 16 to the dealer’s 10 puts you squarely between a rock and a hard place.
But you have a way out of this dilemma. Probabilities. Statistically speaking, the answer to this blackjack dilemma is easy. You hit. The likelihood of drawing a card below 6 (remember anything at or above 6 will cause you to bust) is just slightly higher than the odds of beating the dealer with your 16. Even so, the decision is fraught with angst. Watch any new blackjack player in this situation and you will see the anxiety wash across his face. Someone at the table notices and helpfully advises, “Take the card, idiot!” But the advice is immediately rejected with a knee-jerk, “But I’ll bust.” So, our novice blackjack player sits agonizing over the right move, garnering the glares of the other patrons at the table waiting for their turns. Finally, he whispers sheepishly “stand” and the dealer turns her attention to the next player in line.
Our blackjack player is right of course – in part. He probably would have busted if he had hit. But his decision to stand made things even worse. His poor decision was prompted because, unwittingly, he had framed his decision as a choice between good and bad. If the high risk of busting is bad, then standing must be good. If framed differently, however, as a choice between two bad options rather than one good and one bad option, his decision would have been easier. Sometimes you’ve just been dealt a tough hand. But when you frame the options properly you are better able to use the data available to you to choose the “less bad” of the two.
A choice between two bad alternatives
We submit that the vehicle pursuit question presents a similar illustration of a choice between two bad alternatives. If you pursue, you are putting innocent people at great risk. If you don’t pursue, you are possibly letting a criminal go free.  You are between a rock and a hard place. You have a 16 to the dealer’s 10.
Are those faced with these difficult decisions destined for a life mired in indecision? Will police leaders stop making tough choices? No. By reframing seeming win/lose dilemmas as searches for the lesser of two evils rather than the good versus the bad, a dilemma becomes easier to manage. Indeed, some dilemmas are quickly revealed as not dilemmas at all.
When you take the emotion out of the equation, the pursuit/no pursuit debate is no more a real dilemma than the hit/stand question is at the blackjack table. The data, while admittedly not unassailable, strongly favor a restrictive pursuit policy.  Most credible research demonstrates that the probability of injuring or killing an innocent person (including the police officer him/herself) is greater than the likelihood of apprehending a dangerous criminal because he/she happened to run a traffic light. In "Police Pursuit Driving: Policy and Research," co-author Alpert and Cynthia Lum  studied the evidence on high-speed pursuits and reported the following:
First, we know that pursuits are dangerous. The very nature of a pursuit is that a suspect flees to get away and at least one police vehicle chases to apprehend. The driving behavior creates varying degrees of risk to the public, the officer and the suspect. We know from our interviews of suspects and other reports that most suspects are “glued” to their rear-view mirrors and are focused on the police car(s) behind them rather than paying sufficient attention to the road ahead. We have learned from analyses of pursuit events that while few suspects will stop after a chase has been started, most will continue to flee until the police terminate the pursuit or they crash.
What the research says
Alpert and Lum’s research showed that approximately 72% of pursuits result in arrest, and 30% of all pursuits result in a crash.  However, most of the arrests occur when a pursued vehicle crashes! They further report there is no compelling evidence that crime rates will increase, or case clearances will decline if agencies adopt restrictive policies.
Historically, most pursuits are initiated by traffic or other minor infractions. While there is evidence that some of the fleeing suspects who are apprehended have criminal records, we do not know anything about those not caught or those who are driving the streets and not stopped. There is no evidence that those who flee are the most serious criminals as many advocates for unfettered pursuits claim. Similarly, research conducted in Orlando shows that restricting pursuits does not lead to a significant increase in suspects fleeing. 
Those are the facts. They are not perceptions, arguments, superstitions, or theories. Those facts, when taken together, strongly argue in favor of a restrictive pursuit policy; just like the odds of drawing 10 strongly argue in favor of hitting when you have a 16 to the dealer’s 10 at the blackjack table. Thinking about the issue this way avoids having to look for a good answer among bad alternatives. All we are doing here is looking for the “less bad” answer. And just as at the blackjack table, the data tell us which approach that is.
1. To put the debate in context, here are some relevant data estimates:
- 40% of police pursuits end in a crash;
- 20% of which end in traumatic injury;
- 1/3 of the deaths caused by high-speed pursuits are of innocent bystanders;
- 91% of pursuits involve non-violent crimes; usually traffic violations.
See, Alpert G. Police Pursuit: Policies and Training. National Institute of Justice, Research in Brief, May 1997.
2. For those who don’t play the game of blackjack, here’s a brief, and necessarily incomplete, overview of our card players plight here. In the game of blackjack, every player plays against the dealer – not against the other players. The dealer starts with two cards, the top card face-up, revealing its value to the other players. The bottom card remains face down until the end of the game. The object, put simply, is to collect cards that give you a higher hand than the dealer without going over 21. When it’s your turn, you request a card by saying “hit me,” or you stop receiving cards by saying “stand.” Once each player at the table has requested cards (or not), the dealer then turns over her second card to reveal the total value of her hand. She then requests additional cards to increase the value of her hand without going over 21. The dealer has to abide by certain rules, however. One rule, in most casinos at least, is that the dealer must hit (take another card) if her total hand adds up to 16 or less. Similarly, the dealer has to stand (not take another card) if her card total is 17 or more.
3. Although, it is important to keep in mind that most pursuits are for minor offenses, and most restrictive chase policies cover only non-violent offenses.
4. To be clear, we do not advocate a “no pursuit” policy. For example, certain crimes very well may change the pursuit/no-pursuit cost/benefit calculus. NOPD, for example, has adopted a police pursuit policy that allows pursuits with authorization for violent crimes.
5. Alpert G, Lum C. Police Pursuit Driving: Policy and Research. Springer: New York, 2014.
6. The 40% figure listed in footnote 1 is from the media, which often and sadly has better data than the government or researchers. As there are no national data on pursuits, our best estimate is that 30%-40% of pursuits result in crashes or property damage.
7. Alpert G, Dunham R, Stroshine M. Policing: Continuity and Change. Prospect Heights, IL: Waveland Press, 2006, p. 200.
About the authors
Jonathan Aronie is a graduate of Duke University School of Law. He is based in Washington, DC, where he is a partner and practice group leader at the law firm of Sheppard Mullin, and a co-founder of the firm’s Organization Integrity Group. In 2013, Jonathan was appointed by the U.S. District Court for the Eastern District of Louisiana as the lead monitor over the NOPD Consent Decree.
Dr. Geoff Alpert is a professor at the University of South Carolina. He has served as a member of and has been a subject matter expert to the NOPD Monitoring Team since its appointment by the federal court. Professor Alpert has been researching high-risk police activities for more than 30 years.