Wednesday, December 7, 2011

Interrogation Based False Confessions

Anyone working to reverse a wrongful conviction stands a decent chance of having to deal with a false confession. Since few people can even imagine that they would confess to a crime they did not commit, a confession is deemed compelling evidence of guilty both during and after a trial.

I am working on such a case. Most recently I have drafted a thorough and comprehensive petition for absolute pardon. In that petition, I included a chapter discussing the problem of false confessions in general before turning to the false confession of the petitioner.

I include that chapter below, modified to remove the petitioner's identification. Hopefully, the post will provide useful background information for anyone wrestling with a false confession.

INTERROGATION BASED FALSE CONFESSIONS

There can be no doubt that false confessions occur with distressing frequency. The national Innocence Project reports that in 25% of all DNA exonerations, factually innocent defendants had made incriminating statements, pled guilty, or delivered outright confessions.

In 2003, Rob Warden studied the role of false confessions in Illinois homicide prosecutions since 1970. He found that 33% of the 42 people wrongfully convicted had falsely confessed.

In this post we will consider some of the empirical and experimental evidence that standard police interrogation techniques cause many people to confess to crimes they did not commit. We will briefly discuss three cases of false confession in Virginia that were later corrected by gubernatorial pardon. We will then briefly discuss three experiments that show such false confessions are unfortunately common and easy to extract.

The Norfolk Four
In 1997, four Navy service members confessed to the rape and murder of a woman in Norfolk Virginia. Despite recanting their confessions, and despite being excluded by crime scene DNA, each was found guilty. Three were sentenced to one or more life terms in prison. One was sentenced to 8.5 years.

A few years later, police matched the crime scene DNA against Omar Ballard, who was then serving time for an unrelated offense. Ballard confessed to the rape and murder, gave an accurate description of the physical evidence, and insisted he had acted alone. He added: "them four people that opened their mouths is stupid."

Though none of the Norfolk Four had ever mentioned Ballard, as they freely implicated others, the police changed their theory of the crime. Ballard was the ringleader but didn't want to implicate the others for fear of being labeled a snitch. The others were accomplices who didn't want to implicate Ballard because they feared him.

In 2005, three of the four petitioned Governor Mark Warner for clemency. Governor Warner did not rule on the petition, and Governor Tim Kaine then considered it. In 2009 Governor Kaine granted a conditional pardon to the three petitioners. In 2009, U.S. District Court Judge Richard L. William vacated the murder and rape convictions of Derek Tice, one of the Norfolk Four. In 2011, the United States Court of Appeals for the Fourth Circuit affirmed that ruling.

David Vasquez
In 1984, the nude body of Carolyn Jean Hamm was discovered in her basement of her Arlington, Virginia home. According to the police reports, she had been assaulted and raped before being hanged from a water pipe.

Two witnesses reported they had seen David Vasquez in the area around the time of the murder. The police escorted Vasquez to the police station and interrogated him. They lied to him about finding his fingerprints at the scene. They fed him information about details of the crime and encouraged him to restate them. They yelled at him when his answers did not conform to the details.

Finally, during his third interrogation session, Vasquez told the police he might have helped the victim "move something." Vasquez began speaking of a dream he had that paralleled the crime. The police recorded only the last 8 minutes of Vasquez' three interrogation sections. In that brief recording, David Vasquez recounted his "dream" of rape and murder.

The person who actually raped Carolyn Hamm was a secretor. That means that his blood type could be determined from his bodily fluids, including his semen. Forensic testing thereby excluded Vasquez as the contributor of the semen stains found at the scene and on the vaginal swabs. That fact, however, didn't cause the police to question the confession. They simply modified their theory to include an unidentified accomplice.

The existence of a hypothetical accomplice is so commonly used by police to buttress an otherwise untenable case that the hypothetical accomplice has become known as The Unindicted Co-Ejaculator. As we have seen, a hypothetical accomplice was employed in the case of the Norfolk Four.

Realizing that their client did not stand a chance at trial, given his confession, the Vasquez defense team accepted an Alford plea to avoid the capital murder charge. Vasquez was sentenced to 20 years imprisonment.

While Vasquez was serving his wrongful imprisonment, police investigations tied Terry Timothy Wilson Spencer to four eerily similar murders. Three of the murders had occurred in the months before Hamm's murder; one had been occurred after Vasquez was imprisoned. Though the DNA from the Hamm murder was insufficient to allow testing, given the technology at the time, the DNA from the other four murders proved Spencer to be the rapist in each of those cases.

In 1989, Governor Gerald Baliles pardoned David Vasquez. Regarding her support for the pardon, Commonwealth Attorney Helen Fahey said: "I think everyone involved tried to do the right thing. I think people should know that; that even when the system didn't work, people were doing their job and worked hard. It just didn't work."

Earl Washington, Jr.
No one argues that Earl Washington was a decent person. No one argues, not even Earl Washington himself, that he did not brutally beat and sexually assault 78-year-old Hazel Weeks. Washington broke into her house to steal the .22 caliber handgun she kept on top of her refrigerator. He planned to shoot his brother with that gun. By his own admission, Hazel Weeks was merely an unfortunate and helpless target of opportunity.

It is therefore not surprising that no one argues Earl Washington was a decent person. Earl Washington, however, was not guilty of the crime for which the Commonwealth nearly executed him.

During Washington's initial interrogation, he confessed to his crimes against Hazel Weeks. When he was soon thereafter questioned about a recent unsolved rape of a young woman, he confessed to that crime. When asked about the unsolved murder of Rebecca Lynn Williams, he confessed to that crime as well.

By the end of a series of interrogations, none of which were particularly aggressive, Earl Washington had confessed to three break-ins, two malicious woundings, one attempted rape, two actual rapes, two robberies, a burglary, and capital murder. Earl Washington confessed to every crime about which the police asked.

The police and prosecutor had plenty of reason to be suspicious of Washington's confessions. So inconsistent were they, so devoid of insight into the crimes, most were disallowed before trial. His confession regarding the murder of Rebecca Lynn Williams, however, was allowed to stand.

With respect to her murder, Washington had been unable to provide her race, her height, or the number of times she had been stabbed. He said he had stabbed her once or twice. She had in fact been stabbed 38 times.

Washington did not know where she lived, though she was killed in her own apartment. He did not know that her door was undamaged. He claimed to have kicked it in.

Washington did not know that her two children were with her at the time she was murdered. He did not know she had been raped.

The jury found Earl Washington guilty of capital murder, though there was no forensic or witness testimony to tie him to the crime. It is well established that juries find confessions compelling.

Nine days before his execution, Earl Washington, Jr. was granted a stay when his new defense team petitioned for a state writ of habeas corpus. While the petition was pending, the new defense team discovered a report showing that the Virginia Division of Forensic Science had analyzed semen stains on a blanket from the murder scene. The Commonwealth had dutifully provided that report to the defense, but Washington's trial lawyer had ignored it.

The report showed the semen at the crime scene came from a secretor, from someone with Type A blood. Neither Earl Washington nor the victim's husband had Type A blood. The Commonwealth agreed to subject the biological evidence to DNA testing. That testing confirmed that the stains could not have come from Earl Washington, Jr.

The testing was too late, however, to form the basis of any legal proceeding. Under the law at that time, new evidence had to be presented within 21 days of sentencing. Earl Washington therefore petitioned Governor Douglas Wilder for clemency. Governor Wilder commuted Washington's sentence to life imprisonment. Governor James Gilmore later granted Washington an absolute pardon. After almost eighteen years in prison, most of those spent on death row, Earl Washington finally walked free.

Experimental Evidence of Interrogation Induced False Confessions
In 1996, researchers Saul Kassin and Katherine Kiechel demonstrated that police interrogation techniques frequently cause subjects to falsely confess. More significantly, their experiment demonstrated that people could easily be convinced they are guilty, though they are not. Many of those persuaded of their guilt will confabulate events. They will imagine details to make sense of a false memory or belief.

In the experiment, 75 undergraduate students were asked to perform an alleged reaction-time test. The subject and a test confederate were seated across a table from the experimenter. The confederate read a list of letters and the subject typed them on the keyboard. The subject was specifically warned not to hit the ALT key, because doing so would cause the program to crash and data to be lost. After 60 seconds of testing, the computer seemingly crashed, and the experimenter accused the subject of hitting the ALT key.

Initially, all 75 test subjects denied guilt. Then, in approximately half the cases, the experimenter asked the confederate if she had witnessed the subject hitting the ALT key. By experimental design, the confederate always confirmed that she had witnessed the subject doing so, even though none of the subjects had. In other words, the confederate was there to provide false evidence of guilt.

In all cases, the experimenter then wrote a standardized confession: "I hit the 'ALT' key and caused the program to crash. Data were lost." The experimenter asked the subject to sign the confession, the consequence of which would be a phone call from the principal investigator. If the subject refused, the request was repeated a second time. Overall, 69% of the subjects signed the confession, though none had hit the forbidden key.

These results might be easily dismissed, since none of the subjects faced any serious consequence for falsely confessing to a relatively inconsequential act. Perhaps that is so. The more interesting portion of the test, however, remains to be described. We shall do that now.

The experimenter escorted the subject from the test area, leaving the confederate behind. In the reception area, the experimenter and the subject encountered the next subject to be tested. The next subject to be tested, however, was also a test confederate. The experimenter explained the test session would need to be rescheduled, then left the room to retrieve his appointment calendar. The second confederate then asked the actual test subject "What happened?" The discussions were surreptitiously recorded and later reviewed to see if any of the subjects had come to believe they were indeed responsible for what had happened.

Only unambiguous expressions of guilt were counted. Examples of unambiguous expressions not were "I hit the wrong button and ruined the program" and "I hit a button I wasn't supposed to."

Any expression prefaced with "he said" or "I may have" or "I think" was not counted as a false belief of guilt.

Overall, 28% of the subjects had somehow become wrongly convinced of their guilt during their brief, non-confrontational experience. Even though none of the test subjects had pressed the offending key, and even though none of them had been subjected to the rigors of a police interrogation, 69% of them confessed and 28% of them actually believed themselves to be guilty.

More startling still is the powerful effect of introducing false evidence into the interrogation. In one cohort of test subjects (those subjected to rapid reading of the questions and to false evidence of guilt), 100% of the subjects confessed and 65% of the subjects believed they were indeed guilty.

Clearly, lying to the subjects about the evidence stacked against them can lead to false confessions. It can also cause people to actually believe they committed a crime, though they did not.

One particularly troubling interrogation technique is to subject the suspect to a polygraph exam then tell him he failed though he did not. Even the National Research Council recognizes the risk of such a devious interrogation technique. In 2003, their Committee to Review the Scientific Evidence on the Polygraph published The Polygraph and Lie Detection. From that report, we find:
False confessions are more common than sometimes believed, and standard interrogation techniques designed to elicit confessions -- including the use of false claims that the investigators have definitive evidence of the examinee’s guilt -- do elicit false confessions. There is some evidence that interrogation focused on a false positive polygraph response can lead to false confessions. In one study, 17 percent of respondents who were shown their strong response on a bogus polygraph to a question about a minor theft they did not commit subsequently admitted the theft.
There is still more to be learned from the experiment, beyond the fact that people can be easily prompted to falsely confess, and the fact that many innocent people actually come to believe their guilt. After the experiment was over, the experimenter brought all the subjects back to the lab. He reread the list of letters they had all been asked to type. He asked them if they could reconstruct how or when they hit the ALT key. It was at this point that the experimenters observed evidence of confabulation. Responses showing confabulation were similar to "Yes, here, I hit it with the side of my right hand when you called out the "A."

Overall, 9% of the subjects confabulated details to support their belief that they had caused the computer to crash and data to be lost. The cohort subjected to false evidence of their guilt (and a rapid reading of the letters) confabulated at a much higher rate: 35% of that cohort confabulated details of how they had caused the computer to crash.

Experimental Confirmation of Interrogation Induced False Confessions
In 2003, Robert Horselenberg (and two others) reported that they had effectively replicated the findings just discussed. Horselenberg modified the test such that the experimenter himself claimed to have witnessed the subject hit the forbidden key. In Horselenberg's experiments, all subjects were given false evidence of their guilt.

Horselenberg also increased the cost of a false confession. Anyone who confessed would lose 80% of the fee they were to be paid for their participation in the test.

Despite the increased penalty for confessing, a greater percentage of subjects falsely confessed in Horselenberg's experiment (82%) compared to Kassin's experiment (69%). Also a greater percentage were actually convinced of their guilt after Horselenberg's experiment (42%) compared to those subjected to Kassin's experiment (35%).

Maximization and Minimization Techniques
Lying about incriminating evidence is one many techniques broadly classified as maximization techniques. Other maximization techniques include dismissing the subject's claims of innocence, assuring him instead of his conviction, stressing that no one believes him now and no one will believe him at trial, and implying the penalty will be severe unless he confesses.

At the other end of the interrogation spectrum is a cluster of tactics classified as minimization techniques. These techniques include minimizing the subject's role in the alleged crime, empathizing with him, convincing him the interrogator is interested only in helping him, and implying that the penalty will be minimized if he confesses.

The psychologically based maximization and minimization techniques have replaced the physically based strong-arm techniques of long ago. Implicit threats and promises have now replaced the explicit threats and promises now forbidden by court rulings.

It turns out to be unnecessary to beat a subject, or explicitly threaten him, to extract a confession. Playing with the subject's mind has proved to be just as effective. The new techniques do not seemingly need to be applied with any particular finesse to be effective. Apparently, they need only be applied with persistence and in generous portions.

Maximization and minimization techniques are well established and widely taught. By far the most influential and practical manual at the time was Criminal Interrogation and Confessions, by Inbau, Reid, and Buckley. The book was originally published in 1967. The fourth edition was published in 2001. According to the fourth edition, hundreds of thousands of investigators had already received training in the techniques presented therein. The U.S. Supreme Court has cited the book.

Just this year, in 2011, Allyson Horgan (and three others) reported that they were able to easily extract false confessions using maximization and minimization techniques. In their experiment, undergraduate subjects and confederates were paired, instructed to work together on some tasks and work independently on others. The confederate at some point asked the subject for assistance on a task to be solved independently, thereby giving the subject the opportunity to cheat by helping.

Some subjects cheated, some did not. This experimental arrangement allowed the experimenters to compare the rate of true and false confessions.

In each case, the experimenter accused the subject of cheating, whether they had or not. The experimenter stated that the supervising professor had been notified, was irritated, and might consider the incident to constitute academic dishonesty. The subject was asked to sign a statement admitting their participation in the cheating incident.

With respect to false confessions, the result of Horgan's experiment should no longer be surprising. Of those students subjected to standard maximization and minimization techniques, 42% of those who did not cheat confessed they had.

What is new and revealing about Horgan's experiment is the ratio of false confessions to true confessions. Use of standard maximization and minimization techniques caused one false confession for every two true confessions.

Horgan, however, also subjected a different cohort of subjects to modified maximization and minimization techniques. The modified techniques excluded implicit threats of severe punishment and explicit promises of leniency. The modified minimization relied more heavily on sympathy, flattery, and appeal to conscience; the modified maximization relied more heavily on an unfriendly attitude and a firm belief in the subject's guilt.

Surprisingly, the modified techniques resulted in both a higher rate of true confessions and a lower rate of false confessions. The ratio of true-to-false confessions rose from 2-to-1 to nearly 5-to-1.

Coupled with the earlier experiments of Kassin and Horselenberg, Horgan's experiment adds to a substantial body of work that shows false confessions are easily induced. False confessions are particularly easy to induce when the subjects are presented with false evidence of their guilt, implicitly threatened with dire consequences if they do not confess, and implicitly offered leniency in return for confessions.