Mark Godsey J.D.

Blind Injustice

Is Scott Peterson Innocent? Part Three

Confirmation Bias and the Forensic Sciences

Posted Dec 16, 2018

This is part three in a three-part series on evidence never revealed during the 2004 high-profile trial of Scott Peterson, currently on death row in California for the murder of his pregnant wife Laci and unborn son Connor. In the first article, I addressed how demeanor evidence improperly influenced the case, and in part two I discussed the tunnel vision evident in the police investigation.

One of the leading causes of wrongful conviction is faulty forensic evidence.  While I cannot say whether or not Scott Peterson is innocent, I can say that all three types of forensic evidence presented against Scott to prove his guilt are now understood to be unreliable, if not downright incorrect.

Psychologists studying the influence of bad forensics in wrongful conviction cases have identified confirmation bias as a leading factor.  Confirmation bias or “contextual bias,” occurs when forensic scientists are told by police and prosecutors before they begin their analysis what the “right answer” should be.  Psychologists now understand that when this occurs, it influences how the forensic analyst evaluates the evidence, and can cause them to subconsciously bend their results to confirm what they’ve been told.  This is not intentional.  Rather, it’s how the human mind works. 

Dr. Itiel Dror of London has performed some groundbreaking studies that explain this phenomenon.  For example, in one study, Dror asked fingerprint experts to examine a pair of fingerprints -- one of the suspect, and one from the scene of the crime that the suspect had been convicted of committing.  Dror told the fingerprint experts that the prints were from a case where the fingerprint expert had made a mistake.  The suspect in question was convicted in part because of the fingerprint match, but was later exonerated and freed by DNA testing.  So it was clear that the fingerprint expert had made a mistake in calling the prints a match.  He asked the fingerprint experts to find where the prior fingerprint expert had made mistakes that had led to the wrongful conviction.

What the fingerprint experts didn’t know, however, was that Dror had obtained a pair of fingerprints from a case where the same expert participating in the study had previously testified to a match and had caused the person in question to be convicted.  So he tricked them -- they were actually looking at their own work in an old case where they had found a fingerprint match -- but they were told it was from the case of another fingerprint expert who had made a mistake.

The results were alarming.  Eighty percent of the fingerprint experts changed their answers and said there was no match.  They had testified at a prior trial that the same set of fingerprints were a match and contributed to the conviction, but when told later that they were looking at someone else’s work from a case where a mistake had been made, suddenly the fingerprints no longer matched.

This type of experiment has been replicated time and time again in different forensic fields with similar results.  As I set forth in my book Blind Injustice:  A Former Prosecutor Exposes the Psychology and Politics of Wrongful Convictions, this demonstrates how confirmation bias subconsciously affects output.  Although when I was a prosecutor I routinely told forensic analyist the “right answer” they should expect before they started their work, I had no idea how it could taint the results.  If you would like to know more about how confirmation bias affects forensic analysts, please watch the five-minute video below from leading psychologist Dr. Sherry Nakhaeizadeh:

There were three types of “scientific” evidence presented by experts at Scott’s trial, including a hydrologist’s conclusions regarding water drift and where Laci Peterson’s body would have been dumped in the water, the estimate of Connor’s fetal development to show the date of his death, and the reactions of dogs trained to detect the scent of Laci Peterson when taken to the marina where the prosecution alleged that Scott departed in his boat to dump her body in the bay. As explained below, each of these was problematic.  And in each case, the record is clear that the experts were made aware by the prosecution of the “right answer” before they started.

Peterson family provided upon request
Laci and Scott
Source: Peterson family provided upon request

The hydrologist who testified for the prosecution about the movement of bodies in water explained that if bodies were dumped in the water in one area, they’d be expected to later wash up in a different, specified area based on water movements.  His testimony fit the prosecution’s theory of the case, because he testified that, based on where Laci and baby Connor were later found, Laci’s body would have been dumped in the area where Scott admitted he was fishing on the day Laci disappeared.  However, this “tidal expert” admitted on the stand that he had done no studies, and that he had no expertise, education or practice with respect to the movement of bodies in water. Further, he was well aware of the prosecution’s theory regarding where Scott had allegedly dumped Laci’s body, and where both Laci and baby Connor’s bodies had eventually washed ashore. Therefore, in the terminology of scientific experimentation, he was far from “blind” to the “hypothesis” -- the expected outcome of his analysis. Additionally, the Modesto police had announced Scott’s alibi on TV, and it was widely publicized.  The defense theory was that Laci had been kidnapped; if this were true, the people who murdered Laci knew exactly where to dump her body (in the bay) to ensure the police focus would remain exclusively on Scott.

In the habeas petition filed by Scott’s lawyers that is currently pending in the California Supreme Court, nationally recognized experts who, in contrast to the prosecution’s expert at trial, have actually had extensive training and education in this particular forensic field have now contradicted the conclusions of the expert called by the prosecution at Scott’s trial.

In addition, prosecution expert Dr. Gregory DeVore testified that baby Connor died in utero on Dec. 24, which coincided exactly with the prosecution’s theory of when Scott killed his pregnant wife.  He reached this conclusion based on a formula developed by Dr. Phillippe Jeanty, who “wrote the book” on this discipline, and which involves measuring fetal bones and comparing them to the last known ultrasound to determine when the bones stopped growing.  Before he began his analysis, Dr. DeVore was told that the prosecution believed Connor died on Dec. 24.  So not only did the risk of confirmation bias set in, but worse, to reach the conclusion that would match the prosecution’s theory, Dr. DeVore used the wrong mathematical formula and didn’t measure the correct fetal bones.  According to Dr. Jeanty himself, who has now filed an affidavit in the pending habeas petition, the tibia and fibula, as well as the femur, must be measured. (DeVore only measured the femur).  Based on correct measurements of the correct bones, and using the correct formula, Dr. Jeanty concluded that baby Connor lived well past Dec. 24, and may have been alive as late as Jan. 3, which contradicts the state’s theory and supports the defense’s theory that Laci was kidnapped (see part 2 for discussion of kidnapping evidence).

Wikimedia Commons
Scott Peterson prison photo
Source: Wikimedia Commons

With respect to the dog scent evidence presented at trial, the trailing dog Trimble was reported to have alerted to Laci’s scent near a boat ramp in Berkeley. A trailing dog is different from a tracking dog; the former detects scents in the air, whereas the latter follows a path of physical contact the subject made with a surface. Both of these are distinct from cadaver dogs.  Trimble was a trailing dog who was claimed to have picked up Laci’s scent at the Berkeley marina, where Scott had gone fishing, and where prosecutors argued he’d gone to dispose of Laci’s body. Laci was not alleged to have made contact with the pier or the ground at the marina, however, so having a tracking dog there made little sense.  Importantly, in non-contact vehicle trail scent certification tests, Trimble was wrong 75 percent of the time and ultimately failed his certification test. Additionally, the items used to provide Trimble with Laci’s scent, e.g., Laci’s sunglasses and hairbrush, were susceptible to cross-contamination with Scott’s scent, and Scott is known to have handled Laci’s purse, which contained these items.  And it is undisputed that Scott had been present in those areas at the marina.  Had the dog-handling team followed protocol and conducted a “missing member test” to demonstrate that Scott’s scent was excluded from the items given to the dogs to sniff, this would not have been an issue, but no such test was done.   Nationally renowned dog scent experts have now reviewed the work of the prosecution’s canine experts at trial and have said that the testimony used to convict Scott should be considered “completely unreliable to any expert adequately trained in the field of canine scent detection” and that the “searching protocols employed in this case were virtually guaranteed to produce an unreliable result.

In the end, I do not know if Scott Peterson is innocent or guilty.  What I care about is process.  I study the psychology of criminal investigations and the human flaws that make them go awry.  Innocent or guilty, Scott Peterson’s case is a textbook case for studying these flaws.  According to current understandings of psychology, way too much emphasis was placed on Scott’s demeanor (see part one).  The police developed tunnel vision early on, and lost all objectivity, and this caused them to twist the evidence in their favor, ignore contradictory evidence, and discount potentially good leads (see part two).  And Scott’s conviction was based on forensics that suffered from contextual bias and improper protocols. 

But these problems aren’t inevitable in criminal investigations and can be avoided in the future.  If you think Scott Peterson is innocent, the existence of these problems in this case is extremely problematic, and the real killer(s) are possibly still on the loose.  If you are convinced Scott Peterson is guilty, then you should want to ensure that police and prosecutors are adequately trained to avoid these problems in future cases so that valid convictions are not subject to such easy attacks.

Reporting and writing assistance by Dr. Melissa Berry, Asst. Professor of Psychology at the University of Dayton

Find part one here

Find part two here

More Posts