Skip to main content

Student Projects

For Further Study:

Empirical Evidence in Tort Liability Cases: The Key Ignition Example

Compiled by D. Waytz

In 2009, car thieves in the United States stole nearly 800,000 vehicles, or about 260 per 100,000 inhabitants.  And one major police force estimates that nearly 20% of stolen cars were unlocked, with keys in the ignition.

Undoubtedly, some of these stolen vehicles end up in car crashes, injuring people and damaging property.  Should courts hold liable the owners who left their cars unattended with keys in the ignition?

Almost 40 years ago, in Zinck v. Whelan (1972) a New Jersey court used empirical data to determine that a defendant who had left his keys in an unlocked car might be found liable when a thief stole his car, and hours later, crashed into the plaintiff’s car.   

Zinck and its companion case Hill v. Yaskin (1977) elucidate the problematic use of empirical data in key ignition cases, and highlight the difficulty courts face when dealing with the wider issue of foreseeability. 


One way to determine negligence is to ask what a reasonable person would have foreseen happening and how he or she would have acted under the circumstances. 

The elements of a cause of action for negligence in a key ignition case are the same as in a regular negligence case (13 Am. Jur. Proof of Facts, 1991). Those elements are duty, breach, causation, and harm. Causation can be further segmented into factual ("but for") causation and proximate causation.

Foreseeability is a key factor in determining duty, breach, and proximate causation.  In determining duty, a court asks whether the harm caused to the particular plaintiff was reasonably foreseeable by the defendant.  For key ignition cases, the question is whether there was a reasonably foreseeable risk that a thief would steal the defendant’s car and harm a third party plaintiff, so that the defendant owed a duty to exercise reasonable care toward the injured person.

Similarly, breach may be determined by asking whether the defendant acted unreasonably, because a reasonable person would foresee that his or her actions would cause harm.  In key ignition cases, that foreseeable "act" would be a car theft and subsequent accident stemming from the unattended car.

Proximate cause looks at the sequence of events, and asks whether the defendant’s conduct caused the plaintiff’s injury in a "natural and continuous sequence unbroken by any superseding cause."(13 Am Jur. Proof of Facts, 1991). An intervening event (like a car theft) does not break the chain of causation, if that event was reasonably foreseeable by the defendant.

Thus in a key ignition case, a thief stealing a car from the defendant would not break proximate causation if that theft were reasonably foreseeable under the circumstances.
The courts in our case studies used empirical data to determine the issue of foreseeability on all three factors.


Would a reasonable person foresee that leaving keys in the ignition of an unlocked car creates an unreasonable risk of theft, which could lead to mishandling of the stolen car, and then subsequent injury to an innocent bystander?

Case #1: Zinck v. Whelan

In 1970, Patrick Schoudt, a defendant, left his father’s car parked in front of their house for the night.  He left the door unlocked with the key in the ignition.  During the night, teenagers stole the car and drove 50 miles before crossing over the center line and crashing into the plaintiffs' car, severely injuring three people.  The plaintiffs sued both Patrick and his father for negligently leaving the key in the ignition.  At the trial, the judge granted summary judgment for the defendants, finding no liability.

The Appellate Court believed that in determining duty, breach, and proximate cause, a court should review the foreseeability to a reasonable man of the "enhanced hazard" of theft, ensuing mishandling by the thief, and then subsequent death or injury to a bystander (Zinck, page 445).

A previous decision in New Jersey, Saracco v. Lyttle (1951), held it was not foreseeable that leaving keys in the ignition would lead to theft and subsequent destruction or injury to a bystander.  But in Zinck the court took note of available empirical data in making its decision. 

Empirical Data Used as Evidence of Possible Foreseeability

Empirical data used by the Court included federal and state statistics on auto thefts, as well as a Department of Justice survey of convicted car thieves.

The Court noted FBI data showing car thefts had risen 183% over a ten-year period (1960-1970), which was four times the percent increase in vehicle registrations (Zinck, page 28). In 1964, FBI data showed that 42% of auto thefts occurred when the key was in the ignition (Zinck, page 20). And the Court relied on a D.C. court decision showing that the estimated accident rate for stolen cars was 200 times the rate for normal cars (Myers v. Gaither, 1967).

Also, the Court relied on a survey of convicted car thieves, where questionnaires returned by the thieves indicated that 59% of the thefts involved a key in the ignition, and 18% of stolen cars were crashed (Zinck, page 447-448).


The Court believed the data showing a purported increase in car thefts, and an amplified risk of theft where keys were left in the ignition, in addition to a 1967 auto theft prevention campaign, meant that a "reasonably prudent motor car operator can justly be held to an awareness of the gravamen of the foregoing data."(Zinck) [page 447-448]

The Court thus found a prima facie basis for a factual finding of negligence arising from proof that the defendant left the unlocked car on a public street during the night with a key in the ignition.  There was a question for the jury, and summary judgment was not the appropriate decision. 

Case #2: Hill v. Yaskin: Special Circumstances Fortifying Foreseeability 

Hill v. Yaskin was another New Jersey case where the defendant Yaskin parked her car in a parking lot in a high-crime area, leaving her key for the parking attendant to move the car if necessary.  The car was stolen, and William Hill, a police officer, was injured during the chase.  At the trial level, the court refused to follow Zinck, and held that Yaskin could not have reasonably foreseen the theft and subsequent injury to Hill. 

Empirical Data

The Hill Court noted the data used in Zinck, and augmented it with a federal study that showed an accident rate 47 times greater for stolen cars as opposed to normal cars, and a 24% incidence of theft involving a key left in the ignition or in plain view (U.S. Dept. of Justice, 1975).

Given an "impressive array of statistical information" the Hill Court agreed with Zinck in holding that a factual issue arose as to the foreseeability of this accident.

Hill also noted that special circumstances increased the potential foreseeability of the accident, notably that the lot was in a high-crime area where thefts would be more likely.


Validity of the Data

The empirical data found in the above studies were critical to both the Zinck and Hill holdings.  In using these data, the courts may have been influenced by a 1969 law review article encouraging the use of empirical data in deciding key ignition cases.  In "An Exercise Based upon Empirical Data: Liability for Harm Caused by Stolen Automobiles," Professor Cornelius Peck decried the use of reasonable assumptions in determining foreseeability in key ignition cases, and encouraged the use of statistics from the very studies cited in Zinck and Hill (Peck, 1969).  Professor Peck believed that data showed an increased prevalence of a risk associated with leaving a key in the ignition, and that should make duty, breach, and proximate cause more foreseeable to the reasonable person. 

But Professor Peck noted the data itself might be methodologically unsound.  For example, the survey of convicted car thieves did not include any information about its methods or its non-response rate.  And the survey questioned only convicted car thieves; Professor Peck surmised that those convicted might have been involved in accidents at a higher rate.  The data from the Unified Crime Reports vary widely, do not state methods, do not control for locality, and are based on a chain of interviews from victim to local police to FBI representatives.  And Peck also notes that the data on key ignition thefts are somewhat useless if one does not have statistics on how often keys are left in the ignition without the car being stolen.  Nonetheless, Peck believed the data showed a key in the ignition was a substantial factor in theft.

Neither of our cases discussed the flaws in these studies, and the crime statistics were taken as exact indicators of car theft in the United States.  In essence, the courts believed that the statistics showed increased risk, which in turn should have made the events more foreseeable.  While the data may have been correct, and it seems reasonable to assume a key in the ignition makes theft easier, it is debatable that subsequent theft and injury could be substantially foreseeable, even to a reasonable person.   

An additional problem with using this data to prove foreseeability is the issue of what a reasonable person would know about car theft and keys in the ignition.  At least one California court utilizing empirical data wondered whether a reasonably prudent person would be aware of empirical data showing the enhanced risk of leaving a key in the ignition (Am. Jur. Proof of Facts, 1976).

In total, the use of empirical data in Zinck and Hill was flawed.  Because the statistics were incomplete and varied widely, their entire validity must be called into question.  Additionally, the inability of the courts to show how the mere presence of statistics made the risk more foreseeable harms their analysis.

Alternative Approaches

Perhaps because of these flaws, the New Jersey decisions remain staunchly in the minority for key ignition cases.  Some states decide key ignitions cases under a strict non-liability framework, while others have statutes on point giving notice to drivers that a key in the ignition is unlawful.  

Illinois, for example, has a statue prohibiting a motorist from leaving their keys in the ignition. Here a violation of the statue constitutes a prima facie case of negligence. Other states have similar statutes, but a violation as such will not lead to a finding of negligence.

The majority approach has been one of non-liability, with most courts noting the tenuous nature of duty and causation in these cases.

An Aside: Foreseeability and the Hindsight Bias

Zinck and Hill held that leaving a key in the ignition created a fact question of negligence for the jury, and that special circumstances such as time of day and location should impact the jury’s ultimate finding of liability.

But this holding impacts another area of social science not explored in our key ignition cases, the impact of hindsight bias on findings of foreseeability and liability. 

Professor Baruch Fischoff describes hindsight bias as such: "In hindsight, people consistently exaggerate what could have been anticipated in foresight.  They not only tend to view what has happened as having been inevitable but also to view it as having appeared ‘relatively inevitable’ before it happened.  People believe that others should have been able to anticipate events much better than was actually the case" (Fischhoff, 1975).

Professor Jeffery Rachlinski notes that hindsight bias can have a major impact on determinations of negligence, because a determination of whether a defendant’s behavior was reasonable under the circumstances is made after the fact.  And when an event takes place, the fact-finder will be biased into thinking the event was more predictable, or foreseeable, then it may have been.

In our key ignition cases, the potential impact of hindsight bias is clear.  A fact-finder may determine that a neighborhood was more crime ridden and thus theft was more foreseeable, simply because the theft took place where it did.  Or the fact-finder could have a lower threshold of foreseeability for the simple reason the theft took place after the key was left.

Although Rachlinski notes the difficulty in defeating the hindsight bias, potential defendants in key ignition cases might seek admissibility of hindsight bias evidence to defeat the notion of foreseeability.


The New Jersey key ignition cases highlight the problems with using empirical data in tort liability scenarios.  Due to the difficulty of transferring the impact of statistics to actual elements of negligence, and the court’s inability to thoroughly analyze the data, the Zinck/Hill holding remains an outlier, and the use of empirical data in ignition cases has not taken hold in New Jersey or other jurisdictions.   

Major Cases


  • FBI Unified Crime Reports (1964).
  • Preliminary Study of the Effectiveness of Auto Anti-Theft Devices (Nat'l Inst. of Law Enforcement and Crim. Justice, L.E.A.A., U.S. Dept. of Justice (1975).
  • Hearings on H.R. 15215 before Subcommittee No. 5 of the House Committee on the Judiciary, 90th Congress, 2d Session, at 31-39 (1968).

Further Reading