by Morrison Bonpasse, (email@example.com)
Since the 1989 exoneration of Gary Dotson of his rape conviction in Illinois through DNA testing, over a thousand people have been exonerated from their wrongful convictions. The best source for information about those cases is the newly created National Registry of Exonerations, a joint project of the University of Michigan and Northwestern University Law Schools. These exonerations have correctly shaken our faith in the ability of the criminal justice system to determine truth. In every one of those cases, the truth, as determined "beyond a reasonable doubt" by a judge or jury, was false.
Since the invention of the polygraph in the early 20th century, its reliability as a detector of truth or lies has been vigorously debated and studied. Given the lack of consensus that polygraph results are reliable, they are not permitted, though with many exceptions, to be introduced as evidence of the truth in criminal trials in U.S. courts.
Nonetheless, polygraphs are used extensively by police, lawyers and others in their efforts to determine the truth of people's statements. Faith in the reliability of polygraphs has been enhanced by favorable studies and upon anecdotal evidence where other means have been available to ensure the accuracy of the tested truth. Also, many crimes have been solved by police who announce plans to use polygraphs on witnesses, and the prospect of such a test then sometimes produced a desired confession or changed testimony.
This paper addresses the reliability of polygraphs in predicting the truth which exonerations from wrongful convictions have exposed, before and especially since the advent of DNA testing. Two hundred thirteen (213) exonerations were found, primarily on the Internet, where polygraphs were used with defendants/exonerees, suspects or witnesses, before and after the wrongful convictions. Not all of these cases are listed on the National Registry of Exonerations. However, for those not listed, there is sufficient information for classifying them here as exonerations where the innocence of the defendant/exoneree is affirmed. The 213 cases have been divided into the following six categories where polygraphs of:
1. Exonerees supported innocence. 94 44.1%
Before trial 46 21.6%
After trial 48 22.5%
2. Others supported exoneree innocence. 40 18.7%
Before trial 25 11.7%
After trial 15 7.0%
subtotals 134 62.8%
3. Exonerees had an uncertain result and use. 30 14.1%
Before trial 30 14.1%
After trial 0 0%
4. Others had an uncertain result and use. 18 8.5%
Before trial 17 8.0%
After trial 1 0.5%
subtotals 48 22.6%
5. Exonerees did not support innocence. 23 10.8%
Before trial 23 10.8%
After trial 0 0%
6. Others did not support exoneree innocence. 8 3.7%
Before trial 6 2.8%
After trial 2 0.9%
subtotals 31 14.5%
Of the 94 cases where polygraph examinations of defendants/exonerees supported innocence, 46 examinations were conducted before trial. In those cases, the prosecutors did not believe the validity of the results and obtained convictions from juries which almost always did not know of the results of the polygraph tests, because of the rules of evidence in most courts. For 30 other exonerees, all of whose polygraphs were taken before trial, the results were inconclusive or uncertain. It could be argued that even these results should have been considered by prosecutors as pointing away from the exoneree, given the presumption of innocence, and should have given prosecutors pause before proceeding to trial or attempts to obtain guilty pleas. Thus, in 76 (46+30) instances, out of 213, the prosecutors should have paid more attention to the results of the defendants’ polygraph examinations before trial. Such additional attention could have meant additional investigation, or a request for a second polygraph exam for the defendant/exoneree, or more polygraphs for witnesses. It also could have meant a considerable reduction in suffering by the wrongly convicted people and their families and friends, as well as getting an earlier start of finding and convicting the real criminals.
For 48 exonerees, the polygraph tests affirming their innocence were administered after their wrongful convictions, almost always in prison. Such tests were often used to help assure investigators reviewing a claim of wrongful conviction that they were on the right track.
In 66 (40+18+8) instances, polygraphs were given to others involved in the wrongful conviction cases, such as witnesses or informants. For 40 of those cases, the results supported innocence. Twenty-five of those 40 were administered before trial, but they were apparently ignored or discounted by prosecutors. In 18 cases the results were mixed or uncertain, and eight did not support the exonerees’ innocence. All but one of those 26 (18 + 8) results came from examinations conducted before trial,
Thus, in 134 (94 + 40) or 62.8% of the 213 cases, polygraph results of exonerees and others clearly supported the exonerees' innocence. In another 48 cases, (30 exonerees + 18 witnesses/others), or 22.6%, the results were mixed, inconclusive or uncertain. Therefore, in 182 (134+48) cases (85.4%) the polygraphs reached a result inconsistent with the certainty of defendants/exonerees’ guilt, or, to use the legal terminology, inconsistent with guilt beyond a reasonable doubt.
In 23 cases, exonerees failed their polygraph examinations, and all before trial.
For the 147 (94 + 30 + 23) polygraphs only of exonerees, 94 passed and 23 failed, for a total of 117 with clear results. If the inconclusive/uncertain results are not considered, i.e. the 30, that gives the odds of 4.1:1 or an 80% chance (94/117) of getting a clear correct result for an innocent defendant claiming innocence. If the 30 inconclusives/uncertains are also included, and if the concept is accepted that the only polygraph failure is a clear “deception indicated” failure, the odds of obtaining an acceptable result are 5.4:1 or 84% (94+30)/(94+23+30).
For this article, the more conservative “clear” probability of 80% is used. This rate of accuracy is close to the accuracy rates claimed by the American Polygraph Association (APA). The APA also reports in its Model Policy on Paired Testing that the 2003 National Research Council report, The Polygraph and Lie Detection, found an 86% accuracy rate for polygraphs on single issue testing.
As is described below, there is substantial reason to believe that in many, if not most, of the 23 failed exoneree cases, the polygraphs were defective examinations. In other words, if those exams had been administered by neutral professionals, many, if not most, of the defendants/exonerees would have passed their exams. To the extent that is true, the odds and percentage probabilities of obtaining a favorable result, as will be reported in future versions of this article, will improve when each such defective examination is identified.
A significant difficulty in the researching for this article was the quality of the information about the use of polygraphs, when such use was mentioned at all. Without the full records of polygraph examinations, it is very difficult to determine the quality of the actual exams themselves. Were the examiners properly trained? Where the pre-interviews fairly conducted and were the questions objectively prepared? Even when polygraph examinations were properly conducted, it was not always clear that the reporting of the results was accurate. Sometimes, the police would tell a suspect that s/he failed the polygraph in the hope of generating a confession, even when the examination was passed. This is what happened in the Peter Reilly case, described below.
During the research for this article, the actual reports for defendant/exoneree polygraph exams were obtained for only three cases. The first was the one page report of the 1966 examination in New Jersey for Rubin “Hurricane” Carter, and that record is included in the linked files for him and all of the 213 cases at http://www.bonpasseexonerationservices.com/about.html. The second was also in New Jersey, for Byron Halsey, noted below, who also was said by the police to have failed his exam. The third was a two page report for Erick Daniels, who passed his polygraph in his North Carolina prison in 2003 at the age of 17 after his 2001 conviction for robbery. Each case is linked separately by name on the linked spreadsheet and sorted by the role of the polygraph. Another linked spreadsheet sorts the 213 cases by state. The spreadsheets are attached to this article. As more cases and more information about each case become known to the author, that information will also be updated and linked, and updated versions of this article will be posted. Also, it is hoped that readers of this article who have direct knowledge of one or more of these cases will forward to the author any information and documents which can contribute to the accuracy of this article. Since the first version of this article was posted on the internet and sent to several individuals, the number of cases has grown from 166 cases to the current 213, and the accuracy of the categorizations has improved.
As there were multiple polygraph examinations administered in some of the 213 cases, the actual number of polygraph examinations was approximately 250. These 213 cases came from 34 states, plus the Federal Government, the District of Columbia and Puerto Rico, with the four largest numbers coming from New York (27), Illinois (26), Texas (25), and California (17).
Not included in this article were cases where a polygraph exam played a role, but was not completed. For example, in Illinois, an alternate suspect began to take a polygraph exam in the murder case where Alan Beaman was wrongly convicted in 1995. However, the exam was stopped and canceled due to the suspect’s failure to comply with the polygraph examiner’s procedural instructions. That non-completion of the polygraph session actually contributed to Beaman’s eventual exoneration, but it was a long journey. Because the exam was not completed, it is not included among the 213, because this article is about the accuracy of completed polygraph examinations and not the other implications and uses of polygraphs.
The first known use of a polygraph in a wrongful conviction case was for Jere Snodgrass in 1939 in Michigan. He was accused of passing forged checks and was arrested; the police arranged for a polygraph which he passed. Undeterred, the prosecutors moved ahead with his trial and conviction anyway. The second use of a polygraph in a wrongful conviction case was in Colorado where Loren Hamby was convicted in 1939 of murdering a gas station attendant during a 1937 robbery. He was sentenced to life in prison, but he insisted that he was innocent. His mother had learned of the newly invented polygraph, and asked one of its co-inventors and prominent proponents, Leonard Keeler, to give Loren a polygraph examination in prison. Keeler traveled from Chicago, so the undertaking was a large commitment. Hamby passed the polygraph exam. As often happens with such results, it wasn't the polygraph by itself, which persuaded the prosecutors and the courts of their mistake. Instead, after a period of time, the results helped persuade the chief prosecution witness to repudiate his testimony, and the repudiation was sufficiently credible to persuade the prosecutors. Hamby was released in 1946 and the Colorado legislature awarded him $10,000 compensation in 1947.
The case of Robert Kaupp was the only one of the 213 exoneree polygraph cases to achieve an opinion in the U.S. Supreme Court on the merits of a conviction. Even though Kaupp had passed a polygraph test and even though his co-defendant, Nicholaus Thetford, who had implicated Kaupp, had failed his three (3) polygraph examinations, the Texas Courts upheld Kaupp’s murder conviction in 2001. In 2003 in Kaupp v. Texas, the U.S. Supreme Court reversed Kaupp’s conviction because his confession, at age 17, was wrongly extracted. The Court’s five page unanimous opinion mentions the polygraphs in the body of the opinion, (“Kaupp was cooperative and was permitted to leave, but the brother failed a polygraph examination (his third such failure)”) and in a footnote (“the brother had previously failed three polygraph examinations, while, only two days earlier, Kaupp had voluntarily taken and passed one, in which he denied his involvement.”) These polygraphs must have had some effect on the justices even if they were not explicitly used as justifications for the decision.
Another exoneree Supreme Court case was Randall Dale Adamsv. Texas (448 U.S. 38, 1980), but that was about the constitutionality of the death penalty, rather than a challenge to his wrongful conviction. Thus, there was no mention of the polygraph examinations in his case. Fortunately, Adams’ death sentence was vacated, so he was able to live to see his exoneration in 1989.
In California, Paul Imbler sued his prosecutor, Richard Pachtman, for, among other claims, failing to disclose to his lawyers the results of his polygraph examination. That case made it to the U.S. Supreme Court as Imbler vs. Pachtman in 1975. However, the court affirmed the lower courts’ dismissal of the case on the grounds of prosecutorial immunity. The Court explicitly noted Imbler’s claim “of a lie detector test that had ‘cleared’ Imbler,” but stated later that “but it would have been inadmissible at Imbler's trial and is thus not constitutionally required to be disclosed.” Perhaps now, 38 years later, given the polygraph’s relative success in determining the truth in exoneration cases, and given the requirement that prosecutors are required to provide to defense attorneys all evidence which might lead to evidence admissible in trials, the court would take a different view.
One other Supreme Court-bound polygraph-related exoneree case was about the alleged and egregious prosecutorial misconduct in the wrongful convictions of Terry Harrington and Curtis McGhee in Iowa. However, the case of Pottawattamine County v. McGhee was settled in 2009 for $12 million just before the Supreme Court hearing. McGhee’s and Harrington’s cases are in the “polygraphs of others” category, below. Coincidentally, a major incentive for settling the case on the prosecutor side of the argument was the risk that Supreme Court would overrule the strict immunity from liability for prosecutors rule from Imbler vs. Pachtman.
Examples of each of the categories of the 213 cases are presented below.
Category: Polygraph of exoneree supports exoneree – before conviction
The Peter Reilly case in Connecticut illustrated several uses of polygraphs in police investigations. When his mother was murdered in September, 1973, Reilly was given a polygraph and he passed it. However, the police told him that he failed and used that alleged failure in their successful effort to obtain Reilly's false confession. The 17-year old Reilly reasoned that if a polygraph administered by the police, whom he respected, indicate that he murdered his mother, then he must have done it. This must have happened, he seems to have thought, even if he had no memory of it. Reilly's case fits the sub-category of exonerees passing polygraphs before they were convicted.
In two other cases in this category, defendant/exonerees were told that they failed polygraph examinations that they actually passed. After accepting the case of John Kogut of New York, Centurion Ministry’s polygraph expert, Charles Honts, concluded that Kogut had passed his three polygraph exams, even though the police told Kogut that he had failed.
Similarly, in New Jersey in 1985, Byron Halsey was told by polygraph examiner Peter Brannon that he failed his polygraph test and he was then wrongly convicted of two murders. Halsey was not exonerated until 2007. Since then, he has sued his polygraph examiner and others and Professor Charles Honts was asked by Halsey’s attorneys to evaluate Brannon’s polygraph. Honts found the polygraph to be substandard, and wrote in his 2012 letter,
“In summary concerning the CQT portion of the examination, it is my opinion that there were serious problems with the design and implementation of the Halsey polygraph examination. One of the two comparison questions was at best weak, and was likely inappropriate. The amount of obtained data was less than that required by professional standards. The data that were obtained are inconclusive about Mr. Halsey’s credibility when he answered the relevant questions in both series....
... my evaluation of the entire Halsey polygraph examination indicates that Mr. Halsey did not fail either of the comparison question test series and that the CIT [Concealed Information Test] produced a strong truthful result. It is my opinion that rather than failing this polygraph examination Mr. Halsey actually passed the examination and produced a truthful result.” (emphasis added here.)
In the California child abuse cases involving husband and wife Scott and Brenda Kniffen, they both passed polygraph tests before their trials. If the accuracy of polygraphs is assumed to be 80%, as calculated above for clear results (i.e. excluding the inconclusives), and the corresponding risk of incorrect clear result is 20%, then the likelihood of guilt of two people who both claim innocence and who pass polygraph tests is 4.0% (20% x 20%). A similar result applies in the Indiana murder case of Jerry Watkins where he and his wife passed polygraph tests before his trial. She was his alibi witness. A 4.0% likelihood of guilt does not seem close to guilt beyond a reasonable doubt, but the prosecutors sought and obtained their wrongful convictions nonetheless. The same logic applies when two polygraph examinations are given to a person and the result is the same, as was the case for Jerry Pacek in Pennsylvania in 1959. Only 13 years old, Pacek was prosecuted and convicted of murder despite having passed both examinations.
Perhaps the best known example of a private pre-trial polygraph was the exam for Yusef Salaam, who is one of the “Central Park Five.” His attorney arranged for the polygraph which Salaam passed, but it was not enough to persuade the prosecutors that they were mistaken, at least with regards to his role. As the book and movie of the same name, “The Central Park Five,” show, police and prosecutors also ignored the complete lack of physical evidence linking the “Five” to the assault and rape of the “Central Park jogger.”
Category: Polygraph of exoneree supports exoneree – after conviction
The Dotson and Hamby cases, described above, both fit the category where the polygraph results of the exoneree supported innocence, and the sub-category where the polygraph exam was given after conviction.
A more recent case was that of Marty Tankleff, who was convicted of murdering both of his parents in September, 1988, on the first day of his senior year in high school in New York. His conviction was due primarily to his false confession, which was obtained under duress. After years of appeals, he sought the services of private investigator Jay Salpeter, who agreed to work on Marty’s case after he passed an in-prison polygraph exam.
Centurion Ministries and the Innocence Project often ask their clients, or prospective clients, to take polygraphs for the same reason. Richard Miles, convicted of murder in Texas, took such a polygraph for the Centurion Ministries investigators. His polygraphist was Eric Holden of Richardson, Texas, who also polygraphed exoneree Rickey Dale Thomas.
The Greg Taylor case in North Carolina was another where an attorney had a client polygraphed during the post-conviction appeal process, in order to ensure that the client was telling the truth. In Taylor’s case, he denied any role in the murder for which he was convicted in 1993. What’s not included in the tabulated results here is that Taylor volunteered DNA samples and he offered to take a polygraph immediately after he became a suspect, but the offer was declined. If a suspect volunteers for a polygraph, the police should take one or both of two actions. First, by virtue of the offer along, they should reconsider their suspicion of the volunteer and/or their theory about his/her culpability. Second, if they continue to suspect the volunteer then they should give him/her a polygraph exam or arrange to have one conducted by a neutral third party. In the event of such volunteering by defendant/exonerees, it’s unconscionable to reject both options and continue to prosecute a suspect.
Before the 1997 murder convictions of Christopher Scott and Claude Simmons in Texas, the actual murderer, Alonzo Hardy, had admitted his role to at least one person. However, Hardy’s admissions were not allowed into evidence at the trial. Students at the University of Texas at Arlington Innocence Project took on the case, and Scott and Simmons passed their polygraph examinations. Then, the students interviewed Hardy, who was in prison for an unrelated conviction, and persuaded him to renew his earlier admission of guilt in the case. Then they persuaded him to take a polygraph test to validate that admission. He also passed his test. This was one of only two cases, of the 213, where the actual perpetrator took a post-conviction polygraph to verify an admission of his own guilt. In the other, also in Texas, Martin Kimsey found the man who had committed the robberies for which he was convicted. James Garret was already in prison for other robberies and he admitted to the police his role in the ‘Kimsey’ robbery. Both Garret and Kimsey passed polygraph examinations in support of their claims, and Kimsey was exonerated.
Category: Polygraph of other person supports exoneree
This category includes two types of results. The first is where another person, often a witness, speaks in support of the exoneree’s innocence and passes a polygraph. The second type is where a witness, who is sometimes an inmate informant, provides incriminating information and fails a polygraph.
Ellen Reasonover’s exoneration in 1999 from her 1983 murder conviction is an example of the first sub-category. A prison inmate named Eddie McClenton, came forward to state that the real murderer had confessed to him. Centurion Ministries was working on this case and obtained a polygraph confirmation of McClenton’s statements.
An example of the second sub-category was in the Sonia Jacobs case, where she was sentenced to death in 1976 in Florida for the murder of a policeman. She did not pull a trigger but was with others, and one of the others did pull the trigger. The only person in the group with gunpowder residue on his clothing was Walter Rhodes, who testified against Jacobs and Jesse Tafero, the father of Jacobs’ daughter. The case later unraveled when it was discovered that Rhodes had failed a polygraph exam, but that failure had not been disclosed to Jacobs' and Tafero's attorneys before trial. The prosecutor had justified the plea bargain for Rhodes by falsely affirming that he had passed the polygraph. By the time Jacobs was finally released in 1992, Tafero had been executed. (The mishaps of his bungled electrocution led to the introduction of lethal injections for the Florida death chamber.) If the police had followed the guidance of their own polygraph exam, the wrongful convictions, and the execution of Tafero, would not likely have occurred. Thus, the Jacobs case is in the category where the polygraph examinations of others, i.e. not the exoneree, accurately assessed at least one part of the case, i.e. that the witness pointing to Jacobs, failed the polygraph exam. The case of Sonia Jacobs was one of seven featured in the play, The Exonerated.
The 1975 cases of Clarence Chance and Benny Powell are also examples of where the polygraph correctly challenged the incriminating witness. Here, the allegations of an incriminating informant were apparently doubted after he failed a polygraph exam. He was therefore given a second exam and he failed that, too. The prosecution withheld from the defense the results of the informant’s polygraph failures and Chance and Powell remained in prison until 1992.
Category: Polygraph of exoneree with uncertain result
In New York, Jeff Deskovic was polygraphed during his interrogation for murdering a schoolmate girl in 1989. In 1990, Deskovic requested a polygraph to alleviate suspicion which is what many of the wrongly convicted defendants/exonerees did when faced with the charges against them. As with many other cases, the police used the occasion to obtain a false confession by telling Deskovic that he failed the polygraph. Actually, he seems to have failed only one question during the multiple examinations. Because of that one question, the Deskovic case is classified in the category for exonerees where the results of polygraphs were unclear.
In 2011, polygraph expert Charles Honts evaluated the quality of the Deskovic polygraph in an opinion given in support of Deskovic’s civil lawsuit against the polygraph examiner and other defendants for his wrongful conviction. Honts criticized polygraph examiner Daniel Stephens for using the discredited AEP (Arther’s Examination Procedure). This procedure was discredited even in 1990. Honts found the use of this procedure odd, as Stephens was trained in the more accurate CQT (comparison question test) methods. Honts concluded,
“Issues of technique validity aside, there were serious problems with the technical aspects of the examination conducted by Mr. Stephens. In total those problems are so out of the range of normal practice for the conduct of a valid polygraph examination as to be considered outrageous and contrary to minimally accepted practice in 1990 and before....
Based upon Stephen's polygraph training and experience by 1990 and the undisputed conditions of the polygraph: at least five hours in length, leaving at least some of the sensors on Mr. Deskovic for an extended period of time, failing to use a valid polygraph technique with numerical scoring, and providing Mr. Deskovic coffee and no food for at least six hours, are so out of the normal range for the normal procedures in conducting an actual polygraph examination as to be considered professionally and scientifically outrageous, and in my opinion, the conditions of this examination were simply abusive.
This interaction between Stephens and Deskovic is not consistent with a deception detection CQT, rather this examination is consistent with a guilt presumptive interrogation where the polygraph was used as an evidence ploy to elicit a confession. Moreover, if Mr. Deskovic's testimony concerning, for example the aggressive nature of the questioning, the false reporting of incriminating statements, and the threats and promises made by the Peekskill Police are credited (I make no assessment regarding credibility) the conduct of Stephens would be even more outrageous with regard to accepted practice in 1990.”
All of the results in this category were for polygraphs taken by the defendant/exonerees before trial. In Colorado in 1987, Tim Masters was given a pre-trial polygraph examination by the police and it was inconclusive. After his conviction, a polygraph was arranged by his attorneys and he failed. However, because polygraphs administered before conviction are more important for this article than those performed afterwards, other factors being equal, the Masters case is classified in this article as “exoneree polygraph uncertain.” The Masters case was the only known instance where an exoneree failed a post-conviction polygraph examination.
Another example of the classification issues for this article was the case of Lenell Geter in Texas. He was wrongly convicted of armed robbery in 1982. In one press account, the prosecutor stated that Geter had failed a pre-trial polygraph, but in another, his attorney said that the polygraph was inconclusive. After his release in 1984, Geter took two polygraph examinations in New York, administered by Nat Laurendi, and passed them both. The classification here is “exoneree uncertain,” before trial.
In the case of Anthony Caravella in Florida, who was 15 years old at the time of his polygraph, the polygraph examiner, Tony Fantigrassi, reported that Caravella was truthful in two polygraph responses which acknowledged participation in a rape and murder. However, Fantigrassi reported that “unfortunately,” Caravella’s answer to a third question was “inconclusive,” and Fantigrassi recommended, “Further testing and/or interrogation may be needed to reach a conclusion.” The actual “unfortunate” part of this case was that he and the Broward County Police Dept. did not follow up on the soft recommendation for further testing. What should have reinforced that recommendation was Fantigrassi’s polygraph examination on the same day, January 4, 1984, of Ray Stephen Chappell whom Caravella had named in his confession as a participant in the crime. Chappell was determined by Fantigrassi to be truthful in his denial of involvement in the crime.
Category: Polygraph of other person(s) with uncertain result
In 1990, Antoine Goff and John Tennison were convicted of murder in California primarily because of the testimony of two prosecution eye witnesses. During the investigation, one eye witness, Pauline Maluina, who was 14 at the time of the murder, had recanted her earlier statements and was given a polygraph examination which was said to be inconclusive. Afterwards, the witness was brought together with a cooperating witness to discuss their prospective testimony, and the witness also talked with the prosecutor. At trial, she testified with her pre-recantation version of events, and the polygraph and recantation were never brought to the attention of Goff’s and Tennison’s attorneys.
Category: Polygraph of exoneree with reported failure
There were 23 cases classified as reported polygraph failures by exonerees, and all of them were prior to conviction, and all were by police polygraph examiners.
The best known such failure was the 1966 examination by Sgt. John J. McGuire of Rubin “Hurricane” Carter. The 1966 report stated that McGuire used “standard polygraph procedure” and then stated...
“RESULTS: Subject answered all pertinent questions.
Q. Regarding the shooting at the Lafayette Grill, do you intend to answer
truthfully each question about that?
Q. Did you shoot any of those people last night at the Lafayette Grill?
A. No. Q. Were you present last night at the Lafayette Grill when those people were shot?
Q. Was your car used in that shooting last night at the Lafayette Grill?
Q. Are you deliberately holding back information about that shooting last night
at the Lafayette Grill?
Q. Before the shooting occurred at the Lafayette Grill did you know it was going
CONCLUSIONS: After a careful analysis of the polygraph record of this subject it is
pertinent questions. And was involved in this crime. After the examination and
confronted with the examiner’s opinion, the Subject denied any participation in
Another well-known case of polygraph failures for an exoneree was the Ohio case of Floyd Fay. He was accused of murder and the prosecutor offered to dismiss the charges if Fay would pass a polygraph exam. Fay agreed to take a polygraph and failed it. According to one source, he also failed a second polygraph. By Ohio procedures, his written agreement with the prosecutor made the evidence about the first failed polygraph admissible at his trial in 1978, and he was convicted. Two years later, he was exonerated when his attorney and others uncovered evidence about the real killers, and they were prosecuted. Fay became an articulate spokesperson against the use of polygraphs in criminal trials and his case was commented upon in several journals by polygraph opponents. As his conviction was for a capital offense, Fay also publicly opposed the re-introduction of the death penalty in Ohio after Furman v. Georgia.
One case in this category of failed polygraphs involved the lowest level crime of the 213 cases: embezzlement of the amount of a nightly bank deposit from a hair salon. In 2005, Lisa Hansen left a deposit at a bank in Grand Rapids, Michigan, but the bank said it wasn’t received. Hansen denied stealing the money, but she was fired by her employer. She took a police polygraph in the hope of proving her innocence. However, the polygraph examiner determined that she was “deceptive” and she was later found guilty of embezzlement. According to the National Registry of Exonerations summary of Hansen’s case, “On the advice of her attorney, she elected to plead guilty, but the judge rejected her plea, and Hansen entered a diversion program which cost her $400 and required 40 hours of community service. She completed the program and charges weredismissed in April 2006.” In August of 2006, a bank employee found the missing envelope whereupon the bank apologized and reached an out-of-court compensation settlement with Hansen. Again, the claimed polygraph failure was obviously incorrect.
New Jersey’s third exoneree polygraph failure (after Carter and Halsey) came with one of the most recent polygraph examinations. In 2006, Emmanuel Mervilus was arrested in New Jersey for armed robbery and aggravated assault. He volunteered to take a polygraph in order to prove his innocence. He agreed that the results could be used at a subsequent trial, if any. When told that he failed the exam, Mervilus asked for a second exam, but the request was declined. Lieutenant John Kaminskas of the Union County Prosecutor Office testified at Mervilus’s trial about the accuracy of polygraphs and Mervilus’ failure. A New Jersey court reversed Mervilus’s conviction primarily on the grounds that the polygraph results were relied upon excessively in the trial. The court noted that Kaminskas “testified that he had never encountered a situation in which he had opined that ‘someone was . . . showing signs of deception, and [it later] came out that they were truthful.’ ” In February, 2013, Mervilus was found not guilty in a retrial.
The most devastating exoneree pre-trial polygraph failure came for James L. Dean who was one of the six defendants in the case of the murder of Helen Wilson of Beatrice, Nebraska. The “Beatrice Six,” as they became known, served a total of 87 years in prison for a murder than none of them was actually involved in. The failed polygraph of James Dean was combined by the police with other techniques to secure his false confession and guilty pleas from him and four of the others in 1989. For this article, Dean is classified as a polygraph failure, but there were two others polygraphs of defendant/exonerees in this case. Kathleen Gonzalez’s polygraph seems to have had an uncertain result and Debra Shelden passed her polygraph. For the other three, Joseph White, Ada JoAnn Taylor and Joseph Winslow their cases are classified with “uncertain polygraphs of others,” because of the mixed results of the polygaphs of Dean, Gonzalez and Shelden. Two of the six, Ada JoAnn Taylor and James L. Dean passed polygraphs during the post-conviction re-investigation in 2008 or 2009. However, the pre-trial polygraph results trump the post-conviction results for the purposes of this article. Only one of the Beatrice Six, Joseph White, went to trial and he was convicted in 1989 on the testimony of two of the Six, who, by that time likely believed their own lies. He was still in prison by the time of his DNA-related exoneration in 2008. The others were released earlier, by virtue of the reduced sentences from their guilty pleas.
It is likely that several, if not most, of the unevaluated 23 “exoneree failure” polygraphs will not withstand careful scrutiny by qualified polygraphists. For example, it was widely reported that three of the “Ford Heights Four” failed their polygraph examinations. If the “non-confirmation of guilt or deception” rate of polygraphs is about 80%, as this article confirms, and, conversely the chance of erroneous determination of guilt is 20%, then it is highly unlikely (8/1000 or .008 or .8%, calculated as .20 x .20 x .20) that all three exonerees in the same case failed bona fide, properly conducted exams. Further, there was no evidence from the media accounts of the 23 that any special effort was made to support the original polygraph conclusions. For example, there were no reported requests by prosecutors or defense attorneys to give second polygraph examinations to them before trial.
Another reason for the presumed weakness of the failure results for the 23 unreviewed polygraphs is that several of the exonerees were likely ill-prepared for their exams and tired.
This discussion of polygraph failures by exonerees did not clearly find that police intentionally distorted any of the 23 polygraph results to produce the failures. That such a distortion is possible was documented in the case of Timothy Brown, who was convicted in Florida in 1993 of murdering a policeman which occurred when he was 14, and to which he falsely confessed. In a 2002 hearing on a Motion for retrial for Timothy Brown, Broward County polygraph examiner Richard Hoffman admitted that his polygraph of an alternate suspect was faked. According to an Associated Press story, Hoffman “testified he was ordered to perform the lie detector test as a ruse, didn't ask required control questions and wrote a misleading report on test subject Andrew Johnson.” The Timothy Brown case is not included in the totals for this article because the test was not a real test and therefore not a valid measurement of the accuracy of polygraphs.
The research for this article was restricted primarily to internet-available information. Sometimes only one news article or web address made a difference in the classification. To illustrate this point, consider the John Kogut, Dennis Halstead and John Restivo cases, where John Kogut was given three polygraph examinations by the police. The police told him that he failed them all, so he began to question his own sense of reality and gave a false confession. That confession, in turn, supported the wrongful convictions of Halstead and Restivo. According to the Centurion Ministries summary of the case, John Kogut actually passed his polygraph exams, but that was the only account of the case on the internet which clarifies that point, and that made all the difference for this study for not just one, but three cases. It is anticipated that several of the reported 23 polygraph examination failures were actually instances of non-deception. It is also anticipated that other classifications for this article will change as more information is obtained about each case.
As sometimes happened in the analysis of these exoneree polygraph cases, there was more than one polygraph and therefore the case could have involved more than one classification, but the research design allowed for only one category for each case. When that happened, the classification closest to the exoneree was primary. In “Hurricane” Carter’s case, there were two other polygraphs. One was of Carter’s co-defendant, John Artis, who also was reported to have failed his polygraph. The third polygraph was a given to a witness named Arthur Bello who was in the Lafayette Grill on the night of the murders and who said that Carter was not inside at the time. Before the second trial of Rubin Carter and John Artis in 1976, Bello was given a polygraph examination by Professor Leonard Harrelson and Bello passed. According to the 1987 Third Circuit U.S. Court of Appeals opinion, the chief detective on the case said that Professor Harrelson’s conclusion was “impossible” and then simply ignored the exculpatory result.
The Anthony Graves case in Texas is another case with multiple polygraphs where the polygraph of the exoneree dominated the category classification. Graves failed his polygraph, and that’s how the case is classified for this article. However, the actual murderer (as determined later), Robert Carter, also failed his polygraph. Without the exoneree polygraph, the case would have been classified as one where a polygraph of another person supported the exoneree’s claim of innocence. As noted above, the quality of Graves’ polygraph remains to be examined.
Of the 23 failed polygraphs of exonerees in this category, one hundred percent (100%) occurred before trial, with polygraphs performed by the police. However, one case requiring clarification was that of Tim Masters who failed, as noted above, a post-conviction polygraph administered at the request of his attorneys. Because he had a pre-trial polygraph, which was inconclusive, his case is placed in this article in the “exoneree exam inconsistent” category as the pre-trial polygraphs have a higher priority. If Tim Masters’ second polygraph is considered for the purpose of this paragraph only, then 95% of the reported failures were conducted by police and occurred before trial.
Category: Polygraph of other person with result not supporting exoneree
In 1994, Tiffany Pritchett was charged in Pennsylvania with murder when her co-defendant, Dameon Isbell, claimed that she, and not he, shot a store attendant during a robbery. Isbell passed a polygraph test which prompted the police to offer to polygraph Pritchett during her trial. After a four hour interrogation session, which was not recorded and during which the police took no notes, Pritchett allegedly confessed. It was Isbell’s testimony and Pritchett’s false confession which convicted her. She was released in 2006 after pleading the equivalent of an “Alford plea” in lieu of a new trial. Isbell was never charged with the murder.
In New York, Lynn DeJac had been convicted in 1994 of strangling her 13-year old daughter and was sentenced to 25 years in prison. Her boyfriend, Dennis Donohue, was initially a suspect in the crime, but he passed a polygraph test and became the primary prosecution witness against DeJac. After DNA testing of the daughter’s rape kit pointed toward Donohue, and after a determination that she died of cocaine poisoning, DeJac was released in 2008. That same year, Donohue was convicted of a 1993 strangulation murder in another case.
The results of the 213 cases involving exonerees and polygraphs show that polygraphs assist in the determination of the truth most of the time. The percentages depend upon how the “inconclusive” tests are used, but the percentage of 80% for a clear correct result is used for the purpose of this discussion. Conversely, there would be a 20% chance of failure. If a person claims to be telling the truth and fails a polygraph, then such a person should be given the opportunity to take another, because the probability of an innocent person failing two tests is 4.0% (.20 x .20). For similar reasons, if a polygraph examination for a person claiming innocence reaches an inconclusive result, that person should be offered the opportunity to take another examination.
During the research for this article, only one case, but not involving a wrongful conviction, was found where a person took two polygraphs, with clearly different results, i.e. from pass to fail, or the reverse. In Arkansas, Ed Owens was accused of child sexual abuse and took a polygraph which found that he was deceptive in his denials. Then he took another polygraph which found no deception. The case was dismissed prior to trial.
The largest number of polygraphs taken by any individual in the cases researched for this article was five, by Earl Mann in the Clarence Elkins case. In 1999, Elkins was convicted in Ohio of murdering Judith Johnson, and of attempted murder and rape of her six year old granddaughter. While wrongful convictions are among the unluckiest events one could imagine, it sometimes happens that such people get lucky. For Elkins, the luck came when the alternate suspect, Earl Mann, was sentenced to Elkins’ prison for another crime and he smoked cigarettes. Elkins collected a discarded butt and sent it to his attorney, and the DNA tests showed a perfect match to the crimes for which Elkins was convicted. Still the prosecutor was not sure that the Elkins conviction was wrongful, so she offered polygraph tests to both men. She was still confident that Mann would pass and Elkins would decline to take a polygraph, or that he would fail if he did. So convinced were the police that Mann was not involved in the Johnson murder that when he failed the first polygraph test, they gave him another, and then another, and another, and another for a total of five failures over 10 hours over three days. At a 20% chance of a false reading, the chance that Mann was innocent after failing five tests is .032% (.205). Elkins was released shortly afterwards, in 2005, without the prosecutor even needing to have a polygraph exam for him. The extent to which the prosecutor relied upon Mann’s polygraph results vis a vis the DNA tests is not known. The point here is that the polygraph examinations were accurate about Mann’s deception.
In the cases of 213 exonerees, polygraphs examinations pointed toward the correct result at least 80% of the time. This is still not accurate enough to warrant their use in criminal trials as evidence of guilt or innocence, but this accuracy level is sufficient for investigators and lawyers to use the results of polygraph examinations to more aggressively search for and uncover the truth. One way to further that search for truth is to give more polygraph examinations.
As additional information surfaces about these 213 cases, and especially the claimed 23 polygraph failures by exonerees, and 30 uncertain results for exoneree polygraphs, it is expected that the reliability percentage will increase. Updates of this article will be posted online.
The positive results of this study should send a message to police, prosecutors, defense attorneys, polygraph examiners and the general public that properly administered polygraphs are useful to the pursuit of truth. These results also support the following recommendations:
1. Polygraphs should be encouraged for defendants claiming innocence and for witnesses whose testimony supports that innocence as well as for witnesses whose expected testimony points toward guilt.
2. Investigative polygraphs should be administered by neutral polygraph examiners, instead of using police polygraph examiners. This recommendation is consistent with the widely considered recommendation that crime labs be set apart from police and prosecutor organizations. A backup recommendation is that if an initial polygraph is given by police to a suspect and that person is found to be deceptive, or the examination is inconclusive, AND the person insists on innocence, then the suspect should be offered the opportunity to have a second polygraph by a neutral person. A similar offer should be made for initial police polygraphs of witnesses or informants where the polygraph examination finds deception or is inconclusive.
3. Police, prosecutors and defense attorneys should pay more attention to the results of polygraphs. Every polygraph examination result which is partially or completely inconsistent with police and prosecutors’ theories of a case should give police and prosecutors pause and should prompt a request for a second polygraph test by a neutral examiner. As noted in this article, such inconsistency can work in two ways. First, such inconsistency would appear when a witness or suspect is found to be truthful when contradicting the police theory of a case, or when the polygraph results are inconclusive. Second, the inconsistency can appear when a witness who supports the police theory is found to be deceptive or the examination result is inconclusive.
The 1987 cases of Gordon Steidl and Herbert Whitlock illustrate the importance of this recommendation for re-testing. These co-defendants were convicted in substantial part upon the testimony of Derrell Herrington. During the investigation, Herrington failed a polygraph, so the cases are classified as “polygraphs supporting exonerees.” The polygraph examiner recommended a second test, but the investigators ignored the examination failure and the recommendation; and proceeded to trial with Herrington’s testimony. The fact of his polygraph failure was not made known to defense attorneys
Given the powerful odds noted above of the accuracy of results when two tests are given, it should be considered unethical and unprofessional to proceed in a case in denial of a double polygraph result without a thorough and open re-evaluation of a prosecution or defense. When prosecutors proceed to trial and a conviction in spite of a claim of innocence supported by a doubly passed polygraph, they bear a special burden to justify their actions.
4. Claims of innocence by inmates who have passed, or at least not failed, polygraph examinations should be given more careful consideration. In addition, all inmates who claim innocence (See www.registryofclaimsofinnocence.org) should be offered the opportunity to take a polygraph exam. While searching for exonerees and their cases which involved polygraphs, 40 cases were found where current inmates have passed polygraphs, but their claims of innocence languish. Most of those polygraphs were conducted prior to trials, but the inmates were convicted nonetheless, just as were the 46 exonerees identified in this article. A few of the 40 current inmates have passed their polygraphs while in prison, such as Brian Chevalier in New Hampshire in 2010. If the validity of his polygraph exam is not believed, then he should be given the opportunity to take another polygraph exam. Eight more inmates were identified who have taken polygraphs while in prison which have produced inconclusive results. As argued above, such a result is contrary to the expectation that every person convicted of a crime beyond a reasonable doubt should be found by a polygraph to be deceptive when denying culpability for that crime. When a polygraph does not find clear deception, police and prosecutors should pay attention, and a second test should be offered. Also, where two people have offered different versions of the same event and one is clearly lying, the American Polygraph Assn. Model Policy for Paired Testing could be used.
In the New Hampshire State Prison, Chad Evans took a polygraph exam in April 2012 and the results were found to be inconclusive. Evans is a client of the writer of this article and he has requested a second test, to be financed by his own supporters. The prison warden and his superiors have, so far, denied him such a test. In the research for this article, only one case was found where a prison refused to permit a polygraph exam. That was the case of Alonzo Watts, who was convicted of murder in Pennsylvania. In 2006, a Centurion Ministries volunteer and an investigative reporter were actively working on his case and Watts had agreed to take a polygraph exam, but the prison denied the request. Watts died in prison that year, before he and his supporters had a chance to persuade the prison warden, perhaps via the governor or a court order, to permit the polygraph exam.
5. In 2012, Rob Warden and Ron Frederickson of the Center on Wrongful Convictions recommended that any confessions which police claim to have received after telling a polygraph examinee that s/he failed should not be admissible in court. That recommendation is supported here. Further, as recommended above, when a suspect fails a polygraph examination or whose examination is “inconclusive,” s/he should be offered the opportunity to have a second examination by a neutral polygraph examiner. Of course, all suspects must be advised of their rights to consult an attorney.
6. Every polygraph of a person convicted beyond a reasonable doubt that does not show that person to be deceptive when denying responsibility for the crime should be reviewed. The American Polygraph Association (APA), or American Association of Police Polygraphists, or other group or individuals should fully investigate the 23 cases listed in the linked spreadsheet to this article where exonerees were said to fail, or did fail polygraphs, and the 8 cases where the polygraphs of others were clearly inconsistent with the exonerees’ innocence. From that investigation should come a determination of how many of the 31 (23 + 8) unexplained failures of polygraph examinations of exonerees and witnesses were due to polygraph examiner failure and how many were due to the inherent imprecision of the device and process themselves.
7. The American Polygraph Association could offer to dedicate itself and its members to exonerating the wrongly convicted and to preventing future wrongful convictions. The Association participated in the 1950’s and early 1960’s in Erle Stanley Gardner’s “Court of Last Resort” which evaluated claims of innocence, which sometimes resulted in exoneration. In the 1970’s the APA had a Case Review Committee which provided some polygraph assistance to claimants of innocence. Some State chapters of the APA did similar work into the 1990’s, formally or informally. The recommendation here is that the APA recommit itself to this work of exonerating the wrongly convicted and preventing wrongful convictions. The leading agent of the innocence movement since 1989 has been DNA, but DNA is present in only about 5% of all crimes or major crimes. On the other hand, at least one living human being is present in every single crime, at least when it begins, and therefore, the potential for polygraphs to play a larger role in this work on wrongful convictions is large.
8. The data for this article should continue to be evaluated and updated and posted on the internet.