BIAS
Bias is a preference for or against a person, thing, group, idea, or behavior. Being biased means being “one-sided” and being less willing to consider alternatives. Prejudice and bigotry against certain social or ethnic groups is a form of bias, for example. Although most people like to think of themselves as being unbiased, fair, and impartial, it is nearly impossible to be able to take a completely neutral position. All of our opinions and feelings are based upon our own experiences and we tend to think and act in ways that we were taught as we grew up and which tend to make us more comfortable. Being biased is, in a way, hard-wired into human beings but that does not mean biases cannot be overcome or controlled for. In many cases, people do not even have a conscious realization that they are being biased. Bias is not something that can be turned on or off like a switch and cannot be wished away.
The scientific method is designed to prevent or reduce bias; many other methods take similar precautions. Bias in science can have serious consequences and there are many ways that scientists seek to minimize bias. For example, the United States Food and Drug Administration (FDA) has the responsibility of approving new therapeutic drugs for use in treating disease and other medical indications. When a pharmaceutical company wishes to submit a new drug for FDA approval, a lengthy process is required to ensure that the drug will be safe, will have the effects for which it was designed, and will have identifiable side effects. This process includes extensive testing, first on animals and then on humans. In both of these phases, there are significant points where bias can affect the process of approval. The efficacy of a drug is evaluated in two ways during the testing phases of the drug. There are objective measurements of the biochemical and physiologi cal effects of the drug, such as blood pressure, heart rate, tissue damage, etc. Then there are more subjective observations that are made by the examiners/observers and by the subjects of the testing. This is where bias can cause problems. One of the most common methods used for the evaluation of the effects of new drugs is so-called “double blind” testing. In this process, a population of people who are suffering from the disease or condition for which the new drug is being proposed, is gathered together and divided into two groups. One of the groups will receive the drug and the other will receive a placebo, a formulation that looks exactly like the drug does and contains everything that the drug contains except the active ingredient. Then the effects of the drug and placebo will be observed by the examiners and reported by the subjects. To mitigate the bias that can occur among the patients who think they are getting a drug that will help them, none of the patients know whether they are getting the drug or the placebo. Likewise, the people who are administering the drug and placebo do not know which one they are giving to which patient. This blinding reduces the chances that biases will affect observa tions about the efficacy of the drug. Forensic science is not immune to bias. In fact, forensic science has more opportunities for potential bias because of the close relationship it has with professions that are required to have a particular viewpoint; like police officers, prosecutors, and defense attorneys. Remember that most public laboratories are administered and funded by a unit of government, which has the responsibility to prosecute people for committing crimes. The fact that a forensic scientist is employed by a public laboratory may imply that he or she is biased toward the prosecution and is a “member” of the prosecution “team.” Many forensic scientists who are employed by police agencies are themselves enlisted police officers or civilians sworn as officers who may wear uniforms and carry weapons. These associations may not only give the appearance of bias, but may also actually lead the scientist to become biased toward the prosecution. The topic of bias is diverse, complex, and, at times, contentious. Cognitive bias, poor or bad judgment based on perceptual distortion, illogical interpretations, or irrationality, has been recognized for many years by scientists but its implications for forensic science practice have come to light in recent years, thanks to the research performed by Dr Itiel Dror, a cognitive psychologist, and other researchers. Only two of the many types of bias will be discussed in this chapter. The first is termed contextual bias, where a person is influenced by clues surrounding the topic but not the content itself. For example, when asked to read the letters in Figure 1 (a), people routinely answer “A, B, C.” When asked to read the numbers in Figure1(b), people normally read out, “12, 13, 14.” The interpretation of the middle item depends on whether the viewer has been primed to think of it as a letter or a number. They do this naturally without considering their interpretation has been biased by being asked about letters or numbers. In a typical investigation, the police focus on a particular suspect. Evidence is collected from that individual, along with items from the victim and the scene. All of the focus is on that suspect and may be very difficult for the forensic scientist to not be biased toward associating that individual to the items in the case. In addition, the scientist may be furnished with information pertaining to how the crime occurred (“It was a really vicious crime”), how the suspect became the focus of the investigation (“He’s the victim’s boyfriend, so naturally…”), what other evidence exists that implicates the suspect (“We found a gun in his bedroom”), and other influencing information, little if any that matters to the scientist’s analysis. This situation gives rise to two questions: How much information should an examiner have about the circumstances of the case and how can the focus on the suspect be lessened during the evidence examination process? A number of suggestions have come forth recently to answer both questions. As to the amount of information that the examiner should have when examining the evidence, it has been suggested that only the information needed to perform the examinations should be given out. There have also been suggestions for taking the focus off the suspect. These include giving the examiner several known samples from differ ent but similar sources and not telling the examiner which of these belong to the suspect. Another suggestion involves giving the examiner the unknown evidence first for examination and then only after this is done, are the knowns supplied; in many laboratories, this is currently the process. Without context, no investigation could be conducted nor any forensic examinations be completed—where would one start? However, too much information in favor of one view or another could lead the investigator or scientist astray. Too many view bias as binary, one is or one is not biased. In reality, the situation is more complicated and how much and what information is necessary and sufficient depends on the crime, its context, and what is being asked of the scientist (Figure 2). The other type of bias is confirmation bias. Confirmation bias is the tendency to search for and use only information that supports a belief or hypothesis; any information that conflicts with the belief or hypothesis is discounted or ignored. The most notorious case of confirmation bias in forensic science occurred in the Madrid Bombing case of 2004. In this case, the Madrid police sent a partial fingerprint recovered from a terrorist bombing to the FBI Laboratory and requested that the FBI run the print through their AFIS system to help develop leads. In doing so, the FBI examiner focused upon one of the prints that was returned by the AFIS system. The print belonged to Brandon Mayfield, an Oregon attorney. An FBI fingerprint expert associated the crime scene print to Mayfield. This association was then confirmed by two other FBI examiners, both of whom had access to the report from the first examiner. Even an independent examiner (although a former FBI Laboratory fingerprint examiner) hired by Mayfield’s attorney confirmed the results from the FBI. Every examiner knew that the previous examiner(s) had associated the print to Mayfield and this led them to be biased in favor of also associating it to Mayfield. As it turned out, the Madrid police found the correct suspect and his print was an excellent comparison for the one from the bombing scene; Mayfield was exonerated. Confirmation bias had led to the misidentifications. A subsequent internal investigation by the FBI pointed out the problems of systematic confirmation bias and took steps to mitigate them in their verification protocols. Confirmation bias is especially dangerous at the verification stage of examinations if another qualified scientist is conducting the same type of subjective exam.
FIGURE 1 (a) and (b). When asked to read out the letters in (a), people naturally read out “A, B, C;” when asked to do the same for the numbers in (b), they read “12, 13, 14.” They do not realize that asking for letters or numbers biases them toward an interpretation of the middle item (“B/13”).
FIGURE 2 Rather than being thought of as binary—one is biased or one is not biased—the reality is that context and bias are on a continuum and the amount of information depends on the crime, its circumstances, and what the scientist is being asked to do.
