Information cascades and the prioritisation of suspects

Deep into the night of April 19, 1989, New York City police officers were called to a macabre scene at the north end of Central Park: a twenty-eight- year-old woman namedTrisha Meili had been raped and beaten so brutally that, it was later determined, she lost three-quarters of her blood. (She was comatose for twelve days, and remained in the hospital for several months.) Antron McCray, Korey Wise, Yusef Salaam, Kevin Richardson, and Raymond Santana, five young men from upper Manhattan, aged between fourteen and sixteen, were apprehended by the police, following the first reports of the attacks in the park that night. After hours of police questioning, four of them confessed, on video, to taking part in the attack. In two trials, in 1990, Santana, Wise, Richardson, McCray, and Salaam were convicted of the attack, even though there was no physical evidence tying them to it, only their supposed confessions, which contradicted one another. They were sentenced to terms of between five and fifteen years. The accused came to be known as the Central Park Five.1

On the night of April 19, 1989, there was mayhem in Central Park as roaming groups of youths, perhaps 30 in total, beat and harassed 8 different victims. While the mayhem unfolded, Trisha Meili was being beaten and raped by a single serial offender elsewhere in the park. Police rounded up a large number of the suspected perpetrators. Five of them were charged with the assault on Meili. They became known as the Central Park Five. Although they confessed to crime during police questioning (without their lawyers present), they were exonerated 12 years later when the true perpetrator confessed and his confession was confirmed by DNA evidence. Among other things, the case demonstrates that people sometimes make false confessions under the pressure of police questioning (Russano et al. 2005).

More importantly, false confessions can appear plausible if they fit into a broader frame constructed by investigators and may compound problems caused by confirmation bias or, more generally, tunnel vision.2 Findley (2010) discusses the similar case of Marvin Anderson. Despite vigorously maintaining his innocence, Anderson was convicted in 1982 of robber)', abduction and rape of a 24-year-old woman in Hanover, Virginia. During the assault, the rapist had told the victim that he had a white girlfriend. Marvin Anderson was the only black man that police knew who had a white girlfriend. This fact seemed to be enough to precipitate a series of decisions that led to the arrest and, ultimately, conviction of Anderson even though the evidence pointed more strongly towards another man. It was 20 years before DNA evidence would prove Anderson’s innocence.

Cases like the Central Park Five have inspired a growing literature on the effect of decision-making biases on the investigative decision-making process. The most prominent of the biases that have been examined in the investigative context is confirmation bias and tunnel vision. Tunnel vision leads law enforcement to focus on one offender or one offender type while paying less attention to others. Confirmation bias leads the investigator to gather or weight more heavily evidence that supports a previously held hypothesis. After introducing the concept of confirmation bias and explaining its deeper relationship with the updating of beliefs as new evidence comes to hand, we move on to explore a related but distinct phenomenon called ‘information cascades’. An information cascade occurs when people find it optimal to ignore their own private information. If this happens during a watch listing or suspect prioritisation task, errors could cascade across jurisdictions and through investigative teams or law enforcement agencies.

 
Source
< Prev   CONTENTS   Source   Next >