Research

The 1982 study that found police light bars cost money and saved nobody

The Illinois State Police reviewed 381 of their own crash files and measured the fuel their light bars burned. Then they recommended taking them off every car in the fleet.

What a light bar costs in fuel

Illinois measured it directly. Two runs on the same stretch of interstate, eastbound then westbound to cancel the wind, at a steady 55 mph, with a flow meter reading consumption to a thousandth of a mile per gallon.

Across the fleet that came to $237,387 a year — with the department’s fuel bill for those cars running at $2.6 million in FY1981.

The aerodynamic explanation is the part that makes it real. Canadian research cited in the paper found that a 1976 Plymouth with a roof-mounted light system carried 115 pounds of additional drag at 100 mph, and needed about 31 extra horsepower to hold the same speed as the same car with a bare roof. Broken down: an empty roof rack alone added 10 percent to aerodynamic drag, two beacons another 5, and the siren a further 20 — a 35 percent total increase over a slick roof.

That reframes every top speed figure from the era. Michigan and the LASD both tested with clean roofs and said so; the cars that actually went out on patrol were carrying a permanent 35 percent aerodynamic penalty that nobody was measuring.

The safety question, answered with 381 crash files

The fuel argument was already known in 1982. What makes this paper unusual is that it went after the counter-argument — that light bars keep officers safe — with the department’s own accident records.

Corporal Gerald Leisch reviewed 381 accident files covering every Illinois State Police crash involving the rank of trooper in fiscal 1981. The rank was chosen deliberately: senior ranks get unmarked cars in greater proportion and drive them differently, so restricting the analysis to troopers kept the comparison fair.

The first result is the one that should have caused an argument.

Marked cars were 2.16 times more likely to be in an accident than unmarked ones. Not less safe in some marginal way — more than twice as likely, at a significance level that rules out chance.

The paper is careful about why, and refuses to pick an explanation. It offers three: that drivers of marked cars may develop a false sense of security and drive or park less carefully; that marked cars made up 76 percent of the fleet and may simply accumulate more exposure; and that the troopers assigned unmarked cars skewed towards greater seniority — 33 percent of unmarked-car troopers had twelve to seventeen years of service, the group with the lowest accident rate overall.

That last one is a real confound and the authors say so. But it does not rescue the light bar, because the same pattern held in every district type they examined: in urban, suburban and rural districts alike, unmarked units had fewer accidents.

What the lights actually correlate with

Then the analysis turns to the light bar itself rather than the markings, and finds nothing.

For vehicles involved in accidents there was no statistical relationship between whether a car had a light bar and whether the emergency lights were on. Same result at the California Highway Patrol, from its own separate data — 37.3 percent of slick-top accidents happened with lights on against 39.4 percent for light-bar cars, a difference the paper dismisses outright. Nor was there any relationship between having a light bar and the activity the officer was engaged in, the severity of the crash, the lighting conditions, or the number of lanes on the road.

What did correlate — strongly — was whether the red lights were switched on.

Accidents that happened with the emergency lights running were far more likely to hurt somebody. And 28.7 percent of all lights-on accidents occurred during pursuits, the single largest category, with enforcement stops second at 19.8 percent.

The paper draws the careful conclusion rather than the dramatic one. It does not say officers should stop using red lights. It says that in the situations troopers judge to warrant red lights, the risk is substantially higher — and lists three possible reasons, including, again, a false sense of security provided by the red lights.

The recommendation

Illinois laid out four options: keep the current bars, replace them with more aerodynamic ones, go fully unmarked with covert lighting, or remove the roof equipment but keep the stripes, emblems and numbers.

It recommended the fourth. Grille and deck lights at $77 per car, $77,385 to convert the fleet, against nine-year fuel savings it modelled at $2,497,070 for the Twinsonic cars alone. The car stays marked, stays visible, stays recognisable — and gets better acceleration and a higher top speed into the bargain.

The supporting evidence for keeping it visible came from an earlier study the paper cites: drivers slow by 2.5 to 6.5 mph in response to a black-and-white enforcement vehicle, and a deck-mounted revolving light produced a degree of safety similar to a roof mount. Visibility was doing the work. The height was not.

What happened next — and this part is documented

When this article first went up, it said we did not know whether Illinois acted on the recommendation. That was wrong, and the answer turned out to be in the same federal archive.

They did it. In April 1982, two months after the report, the department put sixty marked patrol cars into service with no roof-mounted lights — chosen at random from the 120 marked vehicles issued that year, with the emergency lighting moved into the grille and onto the rear window ledge. The cars kept their full State Police markings. Illinois called them semi-marked.

The sixty were issued in pairs to officers with comparable rural patrols, so each semi-marked car had a conventional twin working the same roads. A follow-up evaluation by Richard A. Raub, prepared for the Transportation Research Board’s 1985 annual meeting, compared 208 vehicles over the 21 months from April 1982 to January 1984.

Two notes on that table, because precision matters here. The .001 significance attaches to the fuel, cost and accident-rate differences only — the report states that the injury difference was not statistically significant, while also recording that there were no injuries at all to officers driving semi-marked cars. And the report’s own abstract describes the result as “6.5 percent fewer accidents per million vehicle miles”, which does not match its Table 7: 12.6 against 4.4 is a reduction of sixty-five percent. A digit appears to have been lost in the abstract. Where a report’s abstract and its tables disagree, the tables win — and the abstract is the part that gets indexed and quoted downstream.

The accident figure is the one that stops you. Not a marginal improvement — roughly a fifth the accident rate. The ratio between the two types widened from 2.3 to 1 in the 1980 baseline data to 4.7 to 1 in the trial. And across the whole 21 months there were no injuries at all to officers driving semi-marked cars.

Raub checked the obvious objection — that the semi-marked cars might simply have gone to better drivers. The officers’ accident records from 1976 to April 1982, before any of them received the new vehicles, showed no statistically significant difference between the two groups.

The department made it permanent. From 1984 the standard issue was a vehicle without roof-mounted lights, and the policy was extended to every district including urban Cook County. Grille and rear-window lights cost under $100 installed; an aerodynamic light bar could exceed $300. Raub estimated that converting the whole fleet of roughly 1,100 cars would save over $495,000 a year in fuel and accident repair.

One human detail survives in the report. Some officers still preferred the light bar — and because there were plenty of bars left over from traded-in cars, the department simply let them keep one.

So why does every American police car still have one?

That question is now sharper rather than answered, and it is worth being careful here. Illinois ran this trial in rural districts, on 1982 Fords and 1983 Dodges, against 1982 light bars. Each of those matters.

A 1982 Twinsonic is a tall, square, mechanically rotating object. A modern low-profile LED bar is a fraction of the height with no moving parts, and its aerodynamic penalty is nothing like 35 percent — so the fuel half of the argument is a case against 1982 equipment, not against light bars as such.

The accident half is harder to wave away, because it was never about drag. But it is a finding about rural Illinois patrol in the early eighties, and the report itself notes that the lack of strong evidence in urban areas initially held the programme back.

What can be said plainly is this: one American state police force tested the question properly, over two years, with paired officers and a control for prior driving record — and then changed its entire fleet policy on the strength of the answer.

Why it matters to this archive

Two reasons, and the first is practical.

Every performance figure in the Michigan test archive comes from a car with a clean roof. Michigan states it in every book and repeats that in-service performance will be lower. This paper puts a number on how much lower: 35 percent more aerodynamic drag, 31 horsepower, 115 pounds of it at 100 mph. When a 1980 Gran Fury records 124.6 mph, the car on patrol was not doing that.

The second reason is about what these documents are. The police car world runs on received wisdom — that pursuit-rated means something, that a top speed figure describes a car, that the light bar is simply part of what a police car is. In each case somebody measured it and wrote down an answer, and in each case the answer has sat unread in a federal archive for forty years.

This one has been there since 1982.