Thursday, April 5, 2007

Blinded by Science

I don't mean to beat a dead horse, but the adoption of blind procedures in police lineups is one of the most critical reforms being proposed to law enforcement agencies and legislatures as a means of protecting the innocent, and I think warrants a bit more attention. It may seem like common sense to anyone who took Psych 101, but somehow reform advocates still face resistance to one of its most central lessons -- namely, that humans influence each other, and more generally that our biases influence both our perception and behavior, whether we want them to or not. It's not that police officers can't be trusted, but rather that they are human -- just like scientists.

Maybe it would be useful to go back to the beginning, and take a look at what is really meant by "science" in this context, in the first place. According to Levine and Parkinson in Experimental Methods in Psychology, 5 (1993):

[S]cience is not just the organized information that has accumulated from the application of research methods, but includes the research methods themselves. The research methods are arrangements for making and interpreting observations in a manner that minimizes the probability of reaching incorrect conclusions. Thus, science could be defined as both an accumulation of knowledge and an accumulation of research methods that function to limit self-deception by serious observers. It offers some protection against the influence of prior beliefs and the various forms of bias that often are spawned by enthusiasm about an original idea.

This gets right to the point: in order for a procedure to carry with it the legitimacy of science -- and in turn, for any semblance of accuracy to be associated with its results -- it must adhere to certain methodological constraints. Levine and Parkinson found the limitation of self-deception by the observer -- that is, protecting against the insertion of the administrator's bias into the experimental results -- to be so central to the practice of science that they included it in their one-sentence definition. But the authors don't stop there. In the next section in their chapter on the foundations of the scientific method, titled "Objective Methods, Not Objective Scientists," the authors continue (p. 6):
The point has been made that good research design is necessary because personal bias can distort observations and conclusions. Yet it is occasionally implied that scientists are objective individuals, and that it is their scientific detachment and tradition of careful observation that yields the more reliable information on which the society depends. Nothing could be further from the truth.

Which is to say, there's nothing special about scientists that makes their work objective. It's that their work, when done properly, is objective by design. The same should be true for police work, particularly when it deals with suggestible human subjects.
Scientists, including psychologists, like other human beings, can (and usually do) fall in love with their own hunches, ideas, and theories. This is in fact desirable. [...] It frequently takes a great deal of passionate conviction and excitement about an idea to pursue the difficult path of confirmation. [...] Scientists can be expected to be protagonists for their own explanations, and to seek to find data in support of their own theories. On the other hand, this is what makes it so necessary for the scientist to use a set of techniques that can identify useless concepts and incorrect conclusions, and that can force the scientist to recognize unexpected facts. The scientific method and its research procedures perform this function, permitting the scientist to maintain his or her enthusiasm and the bias that this sometimes entails.

Every point in this text on the scientific method applies directly to the context of a properly designed and implemented police lineup procedure. Law enforcement agents are expected to enthusiastically pursue their hunches -- but not at the expense of the truth. This is where police can learn from scientists, when testing the memory of eyewitnesses, who are no less susceptible to suggestion than the test subjects in controlled psychology experiments.
DOUBLE-BLIND PROCEDURES

The usual way of avoiding the problem of experimenters communicating experimental purposes or desired outcomes is the use of double-blind procedures. Double-blind procedures are in effect when neither the subjects nor the experimenters in contact with the subjects are aware of the details of the experiment, the hypotheses [...] Double-blind procedures are possible when the experimenter employs experimental assistants to give instructions to the subjects. The experimenter who designs the research does not have contact with the subjects and is careful to keep the experimental assistants uninformed...

(p. 345) It seems clear why the same protective measures are important in conducting police lineups. We know that eyewitnesses are fallible on their own, and when a shaky memory is coupled with a suggestive, non-blind lineup procedure, the chance of retrieving an accurate reflection of that witness's memory of the perpetrator is small indeed.

And yet police continue to resist the adoption of these basic principles of scientific procedure. When the "Reliable Eyewitness Act" was proposed to the New Mexico legislature earlier this year, which included a provision on this simple principle, law enforcement representatives responded with headlines like "Suspect Lineup Ban Would Make Police Job Harder," and supported their position with the claim that "Current photo lineups and the way they are administered have a long, proven investigative and court record of working extremely well." This is the same "proven record" that has resulted in almost 200 DNA exonerations, 75% of which were the result of faulty eyewitness identifications flowing from the status quo.

But those who have adopted the reforms tend to become advocates themselves, and come to understand that the reform effort is not about distrust of police officers or any suggestion of bad faith on their part. Det. Lt. Ken Patenaude of Northampton, Massachusetts, whose department has been using these reforms for several years now, is one example. In his own words (PDF):
[Blind administration of lineups] is a good procedure that does not intimate the potential of police misconduct but rather eliminates any indavertent cues that the investigator may give to the witness. If the administrator of the lineup is unaware who the suspect is, he cannot unintentionally cue the witness or bolster the witness's confidence by making inappropriate comments about the witness's selection.

I'll close this part of the series with a final passage from Lt. Patenaude:
It's time for change. DNA technology and social science researchers have shown that we need to do a better job of collecting eyewitness evidence in the future. I never want to be that police officer or investigator who was responsible for convicting an innocent person of any crime. Convicting the innocent assures the guilty remain free. It also erodes the public's faith in law enforcement agencies and the criminal justice system as a whole.

Amen to that.

No comments: