Research
Sixty-eight thousand police pursuits a year
In 2012, American police chased 68,000 fleeing vehicles. Over twenty years, pursuit crashes killed more than 7,000 people — about one a day — and two thirds of them were in the car being chased.
The great weakness of the 1990s research into police pursuit was that nobody knew the size of it — and the reason why is its own story. Departments were surveyed about their policies, case files were read in three cities, and a hundred and forty-six people were interviewed in jail — but no one could say how often this was happening across the country.
In May 2017 the Bureau of Justice Statistics answered that, using its own national survey of more than 3,000 agencies and the federal fatality database. The figures are not estimates from advocacy groups on either side. They are the government counting.
The scale
| Agency type | Pursuits in 2012 |
|---|---|
| Local police departments | ~40,000 |
| Sheriffs’ offices | ~18,000 |
| State police and highway patrol | ~10,000 |
| Total | ~68,000 |
Roughly 186 police pursuits a day, every day, somewhere in the United States.
And the rate is not where you would guess. Measured per hundred officers, the smallest jurisdictions chase five times more than the largest: 15 pursuits per 100 officers in local agencies serving under 10,000 residents, against 4 per 100 where the population is 250,000 or more. State police and highway patrol agencies run the highest rate of all at 17 per 100.
A quarter of all local police pursuits happened in departments serving fewer than ten thousand people.
The cost
The fatality figures come from the National Highway Traffic Safety Administration’s crash database rather than from police reporting, which matters — it is an independent count.
Across the twenty years from 1996 to 2015, police pursuits produced more than 6,000 fatal crashes and over 7,000 deaths. An average of 355 a year — about one a day, every year, for two decades. The peak was 2006 and 2007, with more than 400 annually. In 2012 alone, 351 people died.
Who dies is the part that should be better known.
| Who was killed | Share |
|---|---|
| Occupants of the vehicle being pursued | 65% |
| Occupants of an uninvolved vehicle | 29% |
| Bystanders not in a vehicle | 4% |
| Occupants of the pursuing police vehicle | just over 1% |
A third of the people killed in American police pursuits over twenty years — roughly 2,300 deaths — had nothing to do with the pursuit. They were in another car, or standing somewhere.
Does the policy actually change anything?
This is the question the 1990s work could not settle, and the reason this report matters. It sorts agencies by what their written policy says, then counts what they did.
| Written policy | Pursuits per 100 officers |
|---|---|
| Leaves the decision to officer discretion | 17 |
| Restricted by specific criteria | 8 |
| Discourages or prohibits pursuits | 2 |
An eight-fold difference between the loosest policy and the tightest, measured across the same profession in the same year.
The distortion it produces is stark. Agencies with discretionary policies employ just 11 percent of all officers — and account for 19 percent of all pursuits. Agencies with restrictive policies employ 78 percent of officers and account for 69 percent of pursuits.
This is the same finding Omaha produced by accident in the early 1990s, when tightening its policy cut pursuits by 82 percent and loosening it raised them by over 600 percent. Twenty years later, at national scale, it holds.
Where the country actually stands
By January 2013, every state police and highway patrol agency had a written pursuit policy, as did every local police department serving 25,000 or more residents. Among the smallest agencies the gaps remain: 4 percent of local departments and 10 percent of sheriffs’ offices serving fewer than 10,000 residents had no written policy at all.
The direction of travel since the 1990s is measurable. Officers employed by an agency with a restrictive policy rose from 72 percent in 1997 to 78 percent in 2013. Those under a discretionary policy fell from 17 percent to 11. Those in an agency requiring supervisory approval before a pursuit begins rose from 2 percent to 9.
And there is an oddity worth noting. State police agencies are about twice as likely as local ones to leave the decision to the officer — 30 percent discretionary, against 17 percent of sheriffs’ offices and 13 percent of local departments. The agencies working the fastest roads have the loosest rules.
Policy is also not quite destiny. An estimated 486 pursuits took place in agencies with no written policy, and 244 in agencies whose policy discouraged or prohibited them.
What a pursuit is actually like
The report also draws on the IACP’s Pursuits database — 5,568 pursuits from 115 agencies between 2009 and 2013, voluntarily reported, and therefore not nationally representative. With that caveat, it is the closest thing to a description of the typical event.
Most are short. 57 percent lasted three minutes or less and 66 percent covered three miles or less. The televised version is the rare one.
| How it ended | Share |
|---|---|
| The fleeing driver voluntarily stopped | 29% |
| An officer or supervisor discontinued it | 25% |
| A collision | 15% |
| The vehicle crossed into another jurisdiction | 10% |
| The driver escaped, in the vehicle or on foot | about 1 in 6 |
The largest single ending is the driver deciding to stop. The second largest is the police deciding to stop. Together that is more than half of all pursuits ending because somebody chose to end them, not because anybody was caught or crashed.
Which lands exactly where the jail interviews did. When researchers asked 146 people who had fled what would have made them slow down, more than seventy percent said they would have — once the police car was a couple of blocks back and they felt safe. This database says that is precisely how the largest share of pursuits already ends.
The injury numbers, in proportion
Per hundred pursuits in that database there were 2 serious injuries and 10 minor ones. Of the serious injuries, 76 percent were to the suspect, 21 percent to people not involved, and 3 percent to an officer.
About 1 in 200 of those pursuits resulted in a fatality. Set against the national figures — 295 fatal pursuit crashes in 2012 against 68,000 pursuits — the nationwide rate works out lower still.
Both things are true at once, and the honest version of this subject holds them together. Any individual pursuit is very unlikely to kill anybody. There are enough of them that pursuits kill somebody roughly every day, and a third of those killed chose nothing.