In October 2002, residents of the Washington DC area were terrorised by a series of random sniper attacks carried out by two men named John Allen Muhammad and Lee Boyd Malvo. They killed 17 people in total, triggering an all-consuming man hunt lasting two weeks at the peak of their shootings. Early on, leads pointed towards a white van as the main mode of transport for the shooters. As police and news media urged the public to watch out for any white vans, more and more calls would come in of white van sightings at the scene of a recent attack. At roadblocks, police were tasked to focus on stopping white vans over other cars, with the first 2 suspects being arrested in part due to their connection to a white van. In the early hours of October 24th, the two killers were finally arrested. They were driving a blue Chevrolet caprice.

 

While the broader story itself is deeply compelling, I’ve become obsessed with the white van and how it may relate to my own research. Put clearly, am I looking for the non-existent white van? Or perhaps, am I relying too heavily on the existence of the white van?

 

Although this promising lead turned out to be a red herring, the white van did seem like a plausible vehicle for a shooter in this situation. The random arrangement and location of shootings pointed towards someone travelling a lot on the road, maybe sleeping out of the car; someone with a not-so-discrete sniper rifle a windowless van might help conceal. And anyway, vans in general have long been associated with suspicious activity. A parked van may be hiding a highly developed computer set up, ready to hack into all our personal data, or a weird guy in a trench coat ready to entice your child with a bowl of candy. Combine this with the preliminary evidence of van sightings near the crime scene and this makes the white van a solid hypothesis. However, the consequences were devastating, as law enforcement wasted precious time following the wrong leads - even ignoring calls linking an ominous blue caprice to several shootings.

 

There are many reasons why the white van may have been so overemphasised as a lead. You could talk about the very strong public response and pressure to find a perpetrator, or how difficult it would be to look for a killer that shoots his victims at random, from a distance, without leaving any physical or video evidence. But this would not help us find our own white vans. Instead, we can take a closer look at 3 cognitive biases which might have influenced the police’s as well as the public’s decision making and how their interactions might have further strengthened their effect. 

 

The first, most obvious cognitive bias in this story is confirmation bias: white vans are perceived as suspicious. A white van would make sense for this type of crime and there have even been sightings of the white van at a crime scene. It seems so obvious, why even question it? The second bias that might have played a role in this investigation is the Texas sharpshooter fallacy, also related to the clustering illusion. This fallacy describes the practice of seeing patterns in random data. This is in close association with the third bias: the Baader-Meinhof effect – also called frequency fallacy, in which a subject starts to notice something more after becoming aware of it. Through the interaction of both biases, witnesses might have become more aware of white vans at crime scenes after having been made aware of the police’s suspicions. White vans are incredibly common in our vehicular landscape. One of them being present at a crime scene would not have been unlikely in a scenario where the criminals are at the wheel of another car. It seems that the cognitive biases served to both worsen the impact of unbiased preconceptions (suspicious white vans) and provide excessive support to preliminary theories (the white van would be practical for a shooter). The only real indicators that pointed to the white van was the first call of a white van sighting at a crime scene and the theory of the practical vehicle. All in all, very little evidence to justify such a strong focus.

 

In any investigation, there is a certain level of uncertainty which can lead us to make unfounded hypotheses or follow shaky evidence. In research, I find it especially difficult to question how far back your scepticisms should reach when trying to control for cognitive biases - both your own, and that of others. We rely so heavily on previously established knowledge and literature that it would take an insane amount of time to evaluate everything your project is based on. The little time we have often forces us to take theories brought to us by more senior researchers and the literature at face value, even though our experiments might suggest multiple alternatives. We will then often use our lack of experience and fundamental knowledge as an excuse for the data being capricious, when trying to match previously established evidence or build on recent theories (though that may come into play as well).

 

When first hearing about the story, I panicked about where my white van might be hiding, though searching for one seemed like a simultaneously tedious and futile task. Questioning everything we do and everything that’s been done before us is not the solution. Nor is mistrusting other’s research. It is important to keep in mind that any scientific endeavour relies heavily on technique, data type and human interpretation, all of which are imperfect. This depends on us, but also on the scientific consensus of our time. Understanding cognitive biases and their interaction with research may help us in directing our doubt and scepticism to the right subject, instead of, for example, at ourselves.

 

So, do you have a white van in your data? Probably. But where is it?​