*** HE SAID / SHE SAID:***
*** The Paired-testing and Marin Protocol FAQ ***
by Jonathan Marin
Examiners and Certification
*** PAIRED TESTING ***
What exactly is the paired testing concept?
It is a way to reduce the incidence of perjury, that can be applied whenever one of two opposing witnesses must almost certainly be lying. At the request of either party, courts would require both witnesses to be polygraphed, and exclude the testimony of the witness who tests "deceptive" where--and only where--that result is corroborated by a "non-deceptive" result by the opposing witness ("paired results").
Why would two results be better than one?
The concept is based on a well established mathematical principle.
* If you roll one die, the chance of getting a six is 16.66%.
* Roll two, and the chance of getting two sixes is only 2.77%.
If participation is limited to examiners using standardized techniques, who irrespective of other credentials have demonstrated an accuracy rate of at least 85% in a controlled protocol, their probability of getting one wrong result is comparable to rolling one "six"; their probability of getting two wrong results is comparable to rolling two "sixes". Error rates as low or even lower than 2.77% can be confidently predicted.
Paired testing also automatically corrects for the base rate problem. In the real-world of crime solving and spy-catching, the true "base rate" is unknown and unknowable. This tends to confound accuracy statistics and cloud their interpretation. In "He Said / She Said" situations, the base rate is necessarily close to 50 per cent.
Would paired-testing really help solve the problems facing the justice system?
It wouldn't solve them all. But it would be a marked improvement in a number of problem areas.
Applied to criminal cases, paired-testing would reduce the incidence of wrongful
convictions due to perjured testimony by:
* Informants testifying with an expectation of leniency
* Witnesses with an undisclosed interest in the outcome
* Police officers testifying to the voluntariness of confessions and the circumstances surrounding searches and seizures.
Applied to civil cases, they would:
* Reduce the number of groundless lawsuits initiated.
* Reduce the incidence of meritorious suits stymied by perjury-dependent defenses.
* Reduce the load on the courts, thereby speeding justice for meritorious litigants.
* Reduce the incidence of tried cases that are decided incorrectly due to perjured testimony.
* Increase courts' willingness to penalize frivolous litigants and their attorneys.
Will exclusion really prevent perjury more effectively than the threat of prosecution?
Criminal penalties prevent perjury indirectly, by inducing witnesses to perceive that their risks from committing perjury outweigh the benefits they stand to derive from it. In their own best interest they choose to be truthful. Exclusion prevents the false testimony directly, independent of any choice by the witness. The difference is especially important in criminal cases. Prosecution witnesses feel they can lie with impunity, because they know that prosecutors almost never prosecute their own witnesses.
Exclusion would keep a lot of prospective perjury from ever getting near a witness stand. A litigant's prospect of winning declines drastically if his opponent's key witness will be allowed to testify unopposed about the important facts in the case. The high costs of litigation provide a strong incentive against sustaining a case in the face of those odds. Litigants and their attorneys who nevertheless persevered would risk being found frivolous by the court, and burdened with their opponents' legal fees as well as their own. In their own best interest, rational plaintiffs' attorneys will advise their clients to abandon the case and rational defendants' attorneys will advise the offer of a quick and equitable settlement.
Wouldn't the system get bogged down by having to polygraph practically every witness?
No. In many cases where "He Said/ She Said" applies, it is clear when the jury's believing one of the witnesses will probably decide the case. The polygraph testing can be limited to those key witnesses, and to the key points of difference between them. Exclusion should be applied where:
* The case likely hinges on who the jury believes.
* The nature of the transaction makes it unlikely that either party could be honestly mistaken about what happened.
Common examples in criminal cases would be where a defendant disputes a cellmate's claim that he confessed a crime to him, or denies a police officer's assertion that weapons or drugs in evidence were found on his person. There are many comparable situations in civil litigation.
Why wouldn't paired testing apply when 2 witnesses on the same side of a case corroborate each other?
It would. When two witnesses on the same side of a case corroborate each others' testimony, and both test deceptive, the testimony of both should be excluded. An example would be eyewitness testimony, where the eyewitness testifies to the identification, and the police officer testifies that the ID wasn't compromised as by a prior "show-up" or photo array. The eyewitness could be mistaken as to the ID, yet feel sure of the ID despite the lapse in procedure and show no deception if questioned about the ID itself. Questioned about procedural lapses, however, the witness and the officer would both show deception most of the time.
The benefit of this application would be significant. Mistaken identity is the single most common cause of wrongful convictions, and police misconduct is frequently responsible. Suppose polygraph accuracy to be as low as 80%.Then the net change from current practice works out overwhelmingly for the better. In cases where either witness passed or tested inconclusive, exclusion would not apply and there would be no net change. But in ~95% of the cases where it would apply, it would exclude tainted testimony.
Can't jurors usually tell who's telling the truth and who's lying?
The ability of average persons to detect deception in others has been extensively studied. The results reported in the peer-reviewed literature are strong and consistent: they cannot. The average person is unable to tell even when a child is lying. A study by Charles Honts and his colleagues at Boise State, published a few years ago in Applied Developmental Psychology, found that people were no better than chance at detecting lies told by 7 year olds during an interview with a person whom the children believed to be a police officer.
What if both witnesses are lying?
Each witness is on notice as to the risk. The procedure creates a strong incentive for the party who will be less damaged by the truth to tell it.
What if one of the paired witnesses is honestly mistaken?
Where one party is truthful and correct as to the facts, and the other is honestly mistaken, exclusion will be applied against one or the other with a frequency equal to the error rate. The statistical protection that paired testing affords the truthful party is therefore reduced to the extent of the possibility that witnesses could be honestly mistaken. Judges will exercise reasonable discretion to grant or deny requests for paired testing based on their assessment of how likely witnesses are to be honestly mistaken about a disputed fact.
Paired testing would be expensive. Who will have to pay the costs?
The costs will be mostly or entirely offset by the savings it would generate. A lot of prospective perjury will never even get near a witness stand. The judicial system will realize enormous savings not only of judges' time, but of all court personnel -- law clerks, court stenographers, bailiffs, etc. Honest litigants will be spared much of the cost of every aspect of litigation -- meetings with their attorneys, drafting and responding to interrogatories, depositions, pre-trial conferences, and trials.
A litigant's likelihood of winning declines drastically if his opponent's key witness will be allowed to testify unopposed about the important facts in the case. The high costs of litigation provide a strong incentive against sustaining a case in the face of those odds. Litigants who nevertheless persevered risk being found frivolous by the court, and burdened with their opponents' legal fees as well as their own. In their own best interest, rational plaintiffs' attorneys will advise their clients to abandon their case, and rational defendants' attorneys will advise the offer of a quick and equitable settlement.
Can't lawyers be disbarred for using perjured testimony?
Yes they can, but they aren't. The rules are on the books, but have little practical effect. Enforcement against attorneys is difficult. It requires establishing that an attorney actually knew a witness was going to lie, which almost invariably means relying on the word of the witness who lied.
Can't people go to prison if they commit perjury?
Yes, but . . .
The laws are on the books, but prosecutions are rare. Prosecutors have to prove beyond a reasonable doubt both that the witness's statements were false, and that the witness knew they were false. Given the stake that litigants have in the outcome of their cases, it isn't surprising that the threat of punishment hasn't worked well to deter perjury. Exclusion of deceptive witnesses would supplement the deterrent effect of existing perjury statutes with direct, active prevention.
Why would special certification be required?
The statistical justification of the "He Said / She Said" procedure rests on the individual examinations' probability of accuracy. Courts' and litigants' confidence in the procedure requires that there be solid ground for believing those probabilities in every case. That confidence can only come from a results-based licensing process. The very wide range of accuracies documented by peer-reviewed studies indicates wide variation in the capabilities of techniques and in the skill-level of examiners. The NAS study reported a median accuracy of about 86% for a large sample of examiners.
The procedure cannot depend on the average for the profession. Courts must decide particular cases, and rely on the particular examiners in each case. All that is known is that many examiners are competent at the level of accuracy necessary for the proposal to work well, and many others are not. Garbage in, garbage out. A separate, results-based licence must therefore be required. Examiners applying for the license should be required to demonstrate a high accuracy (~85%) in a controlled procedure similar to the protocols of peer reviewed laboratory studies.
As long as the scientific basis for CQT polygraph remains controversial, empirical results -- actual performance -- constitute the only basis for assessing efficacy. Published studies show accuracy rates ranging from nearly perfect down to little better than chance. The high accuracy predicted for paired-test exclusion only holds if both examiners in the pair have personally shown that they are capable of achieving the level of accuracy that underlies the calculations.
Professor Charles Honts of Boise State University has written that any ordinary intelligent person, given a minimum of training, can achieve a high rate of accuracy by adhering to standardized procedures. He may well be right . (The NAS report shows the median for the population of examiners whose proficiency is 86% or better to be 91%.) If so, then the requirement of proof of each individual examiner's proficiency might seem superfluous. Individual qualification is necessary because subtle departures from standard procedure could have large effects. Judges do not have the expertise to detect such departures. It would not do to have experts testifying in every case, reviewing minutia from the videotapes and offering conflicting opinions about whether this or that bit should be deemed a "departure".
If polygraphs are junk in some situations so that we don't admit them, how can we rely so heavily on two such tests? It will be hard to get the legal system to swallow that.
Polygraphs are junk perhaps half the time. There is at present no systematic way to tell which half. The recent NAS report analyzed the accuracy levels actually achieved by polygraphists in more than 50 published studies. The median accuracy was 86%, and the top quartile was at 91%, so approximately half of practicing polygraphists probably are not generating junk. Absent a quantitative assessment of individual polygraphists' accuracy, courts' traditional policy is sensible. One of the key requirements of paired testing is that the participating examiners establish their personal proficiency in order to obtain their Litigation Certificates, so the accuracy of the examiners in any particular case is quantitatively known to the court. If participation is limited to examiners whose proven proficiency is at or above the median, then the arithmetic works nicely; an evidentiary judgment that has a 97%+ likelihood of being correct should no longer be considered junk.
We want to know who is telling the truth in each individual case, not in a run of cases. You are right that if someone fails your test it is highly probable that person is lying, but that is not the way we are supposed to decide cases.
The paired testing idea is designed to reintroduce the idea of "searching for truth" as directly as possible into the way we decide cases. Much of the elaborate apparatus of rules and procedures controlling the admissibility of evidence, the conduct of direct and cross examination, etc., exists to limit material that is unreliable over the run of cases. Hearsay is excluded (or admitted under exceptions) without a determination of the likelihood that the hearsay in question is true or false. The underlying rationale for the rule is that it is not amenable to a procedure - testing by cross examination. No one argues that proffered hearsay actually represents the truth as little as 3% of the time.
Judges have considerable discretion to keep evidence from the jury. Fed.R.Evid 403 (and its counterparts in most states) permits them to bar testimony that is confusing or misleading, or time-wasting. Testimony that has a 97%+ likelihood of being false is almost certainly misleading if believed and a waste of time if it isn't.
Cross examination sometimes helps the jury toward the truth, sometimes not, depending on the preparation of the attorney and of the witness. It is often necessarily oblique in relation to the dispoitive facts, sowing doubt as to them by attacking the witness's credibility in regard to others. Jurors are invited to suppose that people are either truth-tellers or liars, and judge whole testimony accordingly, when in fact almost all people are both truth-tellers and liars, on various subjects in various contexts at various times.
Humans -- and jurors are human -- are not at all good at deciding whether someone is lying or telling the truth. Extensive research has established that in addition to bias factors such as race, clothing, grooming, and diction, people rely on cues that are worse than useless -- fidgeting, eye-contact, vocal projection, and so forth.
The issue is whether to substitute a very low quantified risk of a wrong outcome for a much higher risk which because it is unquantified permits the illusion that it is low. If lawyers and judges took that illusion seriously, they would not call trials "crapshoots".
Maybe there is a clash between scientific culture and legal culture on this - but we don't quantify things like proof beyond a reasonable doubt (more than 97% certain, for example), so it would be hard to convince the legal system to go with your proposal.
1. Jurors estimate probilities, in a vague sort of way. Smith seemed a little more credible than Jones. Or a lot more. It is hardly a bad thing to introduce some precision where it is possible to do so. Also, the legal system effectively quantifies probabilities in at least some circumstances. In almost all civil litigation, the plaintiff bears the burden of proof -- greater than 50%. That is also the threshold for the issuance of warrants, and for grand jury indictments.
2. Paired testing does not seek to quantify the ultimate verdict, only one of the elements that goes into it. Much in life is not amenable to quantification and we have to do the best we can without it. That is not a reason to forego its advantages in areas that are amenable to it.
3. All systems have a built-in inertia -- the legal system perhaps more than most. If paired testing is a good idea, it is worth pursuing in the face of some resistance.
It is tough in our system to set up sort of an independent certification procedure for courtroom experts.
The paired-testing polygraphists will rarely if ever testify as experts. The courts will select examiners for particular cases from the roster of Litigation Certificate holders, where the entry for each listed examiner will report his demonstrated proficiency (his rates of false positive, false negative, inconclusive result and reversal upon review.) As the Standard envisions it, the entity issuing certificates would do so under procedures approved by the ASTM, the APA, and/or the AAPP, and the certificates would bear the seal of one or more of those organizations. States which wished to implement alternative or supplementary certification requirements for their courts are of course be free to do so.
I also worry that the credentials of many experts will be terrific (as they often are today) but they will still tend to favor the defense or the prosecution and these experts will get selected by the two sides. I fear this will especially be the case with polygraphs.
When courts currently qualify examiners as experts, and this is also true of many other types of experts, it is based chiefly on their years of experience and the number of examinations they have conducted. (That is a bit like hiring a salesman based on the number of sales calls he's made, without a clue as to the number of sales he has closed.) The experts are selected by the litigants and paid by the litigants who select them.
The standard's certification and selection procedures are unarguably preferable. The NAS report established that standard techniques in the hands of skilled practitioners can be quite accurate -- 25% of the examiners had accuracy levels of 91% or better. Under the Standard, applicants for certificates submit beginning-to-end videotapes of examinations and the charts those examinations produced. This demonstrates their ability to employ a standard technique correctly to produce interpretable charts. They then must score the charts of at least 80 examinations where ground truth is known, to establish their accuracy numbers.
The examiners for each case are selected from the roster by the court, not the attorneys. The beginning-to-end videotapes and the charts are available for "blind" quality control reviews which can be invoked by either of the parties to the litigation and are administered by an entity independent of the attorneys and the court. If the process results in a strong inference of deception by a witness that witness's side pays the examiners, otherwise the invoking side pays the examiners.
I don't think that we can exclude witnesses based on polygraph testing. Sure, there is a good chance or even an overwhelming chance that one of the witnesses is lying and the other is telling the truth. But the witnesses still need to be given a chance to be heard by the factfinder. Maybe the thrust should be to admit the testing results, not to exclude a witness.
The Standard offers courts the option of excluding testimony, or allowing the inferences to be used for impeachment purposes.
The exclusion option doesn't exclude the entire witness, but only his testimony about facts where he has been found deceptive. Given the high probability that the witness is lying, I believe that courts can exclude under Rule 403 and its counterparts in the states. If not, 403 can be modified to suit.
In practice, the impeachment option will really work out to be the "voluntary exclusion" option in almost all cases. Few litigants (or their attorneys) will let their witness testify, knowing that the jury will learn that there is a 97%+ likelihood that
a) the witness's testimony is false, and
b) the witness knows it to be false.
Under those conditions, the chance to be heard by the factfinder is hardly a chance worth having.
Law enforcement uses polygraph extensively. How could their examiners keep their jobs if they don't produce good results?
Overall accuracy frequently isn't their main criterion. Accuracy is not measured or reported at any step, so it isn't even known. Departments' assess their polygraph examiners largely on their ability to elicit admissions.An examiner can have a successful career by eliciting them, irrespective of his ability to consistently produce and interpret meaningful charts. Polygraph schools' curricula devote a considerable portion of the students' time to interrogation technique and eliciting admissions, probably with these real-world criteria for job-success in mind.
What percentage of practicing examiners can achieve the necessary level of accuracy?
No-one knows. The American Polygraph Association does not list demonstration of a candidate's success rate among its requirements for full membership. It lists training, years of experience, and number of examinations administered -- but nothing about a success rate. I have reviewed the CVs and web pages of dozens of examiners, and found not one that reported a success rate. The peer-reviewed studies I have read report their participating examiners' qualifications in terms of their training and the number of prior examinations they have completed, but not their success rate.
I think this may explain the wide disparity of reported accuracy in laboratory and field studies. When a profession defines its qualifications in terms of At Bats, without regard to Batting Average, it should hardly surprise us to learn that quite a few of its practitioners can't hit well, and that some can't hit a lick.
Won't it be expensive to qualify each practitioner individually?
Yes. These procedures would be expensive. Even a study involving 80 subjects - 40 programmed guilty and 40 programmed innocent - would have a rather coarse minimum incremental precision of 2.5%, which in practice would be even coarser due to inconclusives.
As with any valuable credential, the people who want it will bear the cost of obtaining it. The credential will be very valuable indeed to those who earn it. They will gain access to a large new domain, and competitive advantage in existing domains. It would prove to be a license worth having, and the practitioners who qualify for it will quickly recoup their costs.
The American Polygraph Association accredits schools and training programs. Aren't their graduates qualified?
Some are, some aren't. I think that the American Polygraph Association embarrasses itself and does the field a disservice by clinging to claims of 98% accuracy. Just as there are too many studies showing ~85% or ~90% for responsible opponents to ignore, there are too many with accuracy below 60% for APA to ignore without losing credibility.
The main professional requirements for full membership in the APA are graduation from an APA accredited school, and the performance of a specified number of examinations. The accuracy of those examinations is not among the criteria. Even the faculties are questionable. APA accredited schools advertise the apparently impressive qualifications of their faculty, citing their many years of experience with large police departments or federal agencies, and the thousands of examinations they have administered.What they don't say, and presumably don't know, is the percentage of those thousands of tests that yielded a correct result.
The spotty accuracy demonstrated by studies conducted by "highly qualified" examiners becomes readily understandable. Novices learn unmeasured skillsets from masters whose qualification is that they have long experience with them, and then go out into the world and apply them. It is as though a medical school hired practitioners to teach radiology without evidence as to the accuracy of their diagnoses, requiring only that they had read a large number of x-rays, or a golf academy graduated students based on the form of their swing, without regard to where their balls actually went.
Don't polygraph examinations depend on the examiner lying to the subject? What if the subject is resistant or skeptical?
During the pre-test portion of the CQT procedure, the examiner will say things to the subject that may of may not be true, with the object of stimulating heightened physiological responses. The effect doesn't depend on the statements being believed by the subject, but on their focussing his attention, and on their power of suggestion.
In one typical variation, the examiner asks the subject to remember a time when he lied to someone who loved and trusted him, and to remember how uncomfortable that made him feel. He asks him to remember a time that he lied about something important, where there was a great deal at stake, but got caught. He then asks him to recall how he felt when that happened. The purpose is to draw the subjects attention to stressful feelings associated with lying that he has had in the past, and refresh the connection.
The cooperative subject will think about them as requested; the resistant subject may try not to think about them. But it is difficult or impossible to will oneself to not think about something, so the effect is the same for both. Try it yourself. For one minute, starting now, try to NOT think about the Great White Bear that lives at the North Pole. It may be possible for subjects to resist this, but it very unlikely unless they are deaf or unable to understand English.
Later, while calibrating the equipment, the examiner will ask questions such as "Have you thought about what could happen to you if you fail this test?" He will then tell the subject that his physiological responses are very clear, and reassure him that if he just tells the truth he has nothing to worry about. This is a "canned" line - the examiner always says it regardless of the readings on his instruments. Suppose that the subject intends to be deceptive, has read a book on how to "beat" the polygraph, and knows that it is a canned line. The canned statement might be true. It might not. The examiner knows, but the subject does not. The subject's attention has been drawn to the consequences of failure, and to the fact that the examiner alone possesses a crucial bit of knowledge. Knowledge is power; powerlessness is stressful. Uncertainty is stressful. It is not necessary for the subject to believe the examiner's statement. The subject's attention having been drawn to the possibility that it might be true; the combination of uncertainty and powerlessness are then sufficient.
Don't peer-reviewed studies show results that are no better than chance? Does polygraphy really have a scientific basis.
Studies that show accuracies no better than chance merely document the inadequacy of some procedures, some examiners, and some experimental protocols. Polygraphy is not an inseparable "it" that must be accepted or rejected whole.
It is scientifically well established that lying causes stress, and that stress causes measurable responses in the Autonomic Nervous System (ANS). Some opponents of ANS polygraphy argue that anger and fear cause the same ANS responses as stress due to lying, are natural and frequent in the context of PDD exams, and cannot be distinguished by PDD equipment or procedures. Eliminating or accounting for confounding variables is fundamental to scientific method. They argue that since ANS polygraphy cannot do this, it fails to meet an essential prerequisite to being thought of as scientific, voiding any claim to a scientific basis, thus rendering further discussion pointless.
The key word is "cannot". The assertion that examiners cannot distinguish among the possible causes of an ANS response is a falsifiable hypothesis, which has in fact been tested empirically by the high accuracy reported in many studies published in the peer-reviewed literature. The assertion that examiners cannot make the distinctions is falsified by the fact that some examiners have in fact done it. That others cannot does not affect the falsification, but only points to a need for determining which procedures and examiners can discern the differences and which ones cannot.
*** MARIN PROTOCOL ***
What is meant by "strong
"Strong inference" denotes inferences drawn when the conclusions of two or more independent examinations reinforce each other.
When the result of a pair of examinations are both conclusive, how do we know they are both right?
We don't. The key is the linkage between the witnesses' relationship to the facts. Since it is known that one of a pair of opposed witnesses is lying, then if both are found deceptive, or both found non-deceptive, one of the results is wrong. It doesn't matter which one -- no action pursuant to the Protocol will be taken regarding either witness. The only combination of results that can yield a strong inference is that one of them is found deceptive, the other non-deceptive. The inference is strong because for it to be mistaken, both results must be wrong. The same reasoning applies to mutually corroborating witnesses.
Will paired testing always result in excluding the
testimony of a witness?
Paired testing may be implemented in several ways. It may be implemented to permit the use of the examination results for impeachment of witnesses found to be "deceptive" regarding facts where a contradicting witness has tested "non-deceptive" about those facts. Alternatively, it may be implemented to exclude as misleading deceptive witnesses testimony. In all implementations, action is only taken when the testing has yielded a strong inference regarding the truthfulness and deceptiveness of the witnesses.
There is a substantial risk that paired testing will
yield no strong inference. Won't that discourage honest litigants from invoking
Only where the stakes are low. From the honest party point of view there is ~50% chance of a quick win, ~1% risk of a quick loss, ~50% chance of no effect. Paired Testing will either quickly and inexpensively resolve the case, or have no effect on it. In the ~50% of cases which it resolves, the risk of a wrong outcome is very small.
Above a low threshold the decision whether to take the risk is therefore straightforward. The risk taken balances favorably against the reduced risk of ultimately losing the case or being forced by delay and expense into an unfavorable settlement or plea bargain. If paired testing succeeds it wins the case quickly, eliminating exposure to unpredictable open-ended attorneys' fees and other costs. The effect if it fails to produce a strong inference is only to add a relatively small amount -- a few thousand dollars - to the overall cost of the litigation.
The mathematics works if the tests are independent, but paired testing requires subjects to be examined about the very same facts. Doesn't that undermine independence?
It is the examinations themselves that have to be independent, not the particular facts that the subjects are examined about. Two flips of the same fair coin are as much independent trials as flips of two different coins. The paired testing concept works precisely because of the linkage between the subjects' relation to the facts. When a pair of opposing witnesses is examined and both test deceptive or both test non-deceptive, then one of the conclusions is wrong. If one tests deceptive and the other tests non-deceptive, then the conclusions are either both right or both wrong, and the Protocol is structured to support a strong inference that they are both right. The same reasoning applies when a pair of corroborating witnesses is examined. If they both test deceptive, or both test non-deceptive, then the conclusions must either be both right or both wrong, and the probability is high that they are both right.
How often will strong inferences be obtained when the results of both tests are conclusive? How often will they be right?
The examples below suppose paired testing, implemented to exclude testimony from deceptive witnesses. They illustrate numerically the effect the action (if any) that would be taken per 100 pairs of witnesses where a conclusive result is obtained for both subjects, assuming examiners whose proven accuracy is 80%, 85%, and 90% accuracy respectively.
+ means that exclusion increased the likelihood of a just result
- means that exclusion increased the likelihood of an unjust result
= means no exclusion (same as under present system)
a) Supposing accuracy of 80 %:
80 Non-deceptive subjects correctly identified
64 (of deceptives paired to the 80) correctly identified, properly excluded (+)
16 (of deceptives paired to the 80) incorrectly identified, neither subject excluded (=)
20 Non-deceptive subjects, incorrectly identified
4 (of deceptives paired to the 20) incorrectly identified, wrong subject excluded (-)
16 (of deceptives paired to the 20) correctly identified, neither subject excluded (=)
TOTALS (64+, 4-, 32=) or (+) in more than 94% of cases where exclusion applies.
64 cases where deceptive subject correctly excluded, non-deceptive subject allowed
4 cases where deceptive subject erroneously allowed
32 cases where both witnesses would testify as they would now
b) Supposing accuracy of 85 %:
85 Non-deceptive subjects correctly identified
72 (of deceptives paired to the 85) correctly identified, properly excluded (+)
13 (of deceptives paired to the 85) incorrectly identified, neither subject excluded (=)
15 Non-deceptives, incorrectly identified
3 (of deceptives paired to the 20) incorrectly identified, wrong subject excluded (-)
12 (of deceptives paired to the 20) correctly identified, neither subject excluded (=)
TOTALS: (72+, 3-, 25=) or (+) in 96% of cases where exclusion applies.
72 cases where deceptive subject correctly excluded, non-deceptive subject allowed
3 cases where deceptive subject erroneously allowed
25 cases where both witnesses would testify as they would now
c) Supposing accuracy of 90 %:
90 Non-deceptive subjects correctly identified
81 (of deceptives paired to the 90) correctly identified, properly excluded (+)
9 (of deceptives paired to the 90) incorrectly identified, neither subject excluded (=)
10 Non-deceptives, incorrectly identified
1 (of deceptives paired to the 10) incorrectly identified, wrong subject excluded (-)
9 (of deceptives paired to the 10) correctly identified, neither subject excluded (=)
TOTALS (81+,1-,9=) or (+) in more than 98% of cases where exclusion applies.
81 cases where deceptive subject is correctly excluded,
1 case where deceptive subject erroneously allowed, non-deceptive subject excluded
18 cases where both witnesses would testify as they would now
When inconclusive examinations are factored in, what are the odds that testing will yield strong inferences, such that deceptive testimony can actually be excluded or impeached? What are the odds that the testing under the Protocol will get it wrong, and the honest litigant will be barred or impeached?
Supposing that both examiners have exactly the minimum recommended accuracy (86%) and conclusive (80% ) rates, then the distribution of results will be:
0.473 Both examinations conclusive, both correct (Inference correct)
0.0125 Both examinations conclusive, both incorrect (Inference incorrect)
0.276 One examination conclusive and correct, one examination inconclusive (No inference)
0.056 One examination conclusive and incorrect, one exam inconclusive (No inference)
0.154 One examination conclusive and correct, one exam conclusive and incorrect (No inference)
0.04 Both examinations inconclusive (No inference)
Note: The sum of 1.0115, rather than 1.0000, is an effect of rounding during intermediate calculations.
The table shows that where the protocol yields a strong inference, it will
favor an honest witness about 38 times for every time it harms one. The
likelihood of a strong inference from completed examinations is about 50%.
In practice the results will be somewhat more favorable to non-deceptive
witnesses, and unfavorable to deceptive ones. This is due to several factors:
Some subjects will make adverse admissions in the course of their session. Many examiners are skilled interrogators, and this number should be substantial. For methodological reasons these results had to be excluded from the NAS study, and from the Protocol calculations, but as a practical matter they will be significant.
Some examinations will be terminated due to attempts at countermeasures repeated after warning.
Typically, one or both examiners will have qualified well above the bare minimum level. (The NAS report shows the median accuracy for the population of examiners whose proficiency is 86% or better to be 91%).
The recommended minimum accuracy rate is 86%. Shouldn't the minimum rate of "conclusives" be the same?
Perhaps. An examiner's "inconclusives" rate reflects his assessments of charts of completed examinations. Subjects who make adverse admissions, or are found deceptive after being detected and repeatedly warned for attempting countermeasures, are excluded from conclusive-rate as well as accuracy-rate calculations. In practice, results from examinations that did not go to completion will significantly reduce most examiners' "inconclusives" rate.
Jurisdictions are not bound to the Protocol's minimums. Judges will presumably select the available examiners with the best combination of accuracy and conclusiveness.
Isn't perjury still going to happen anyway?
The Protocol is not a panacea for all perjury. It will be effective in the specific circumstances where it applies. It won't eliminate perjury, but will reduce it substantially.
Will prosecutions for perjury be more frequent as a result of the Protocol?
They definitely ought to be. Perjury is rampant but seldom prosecuted, ostensibly because it is hard to prove beyond a reasonable doubt. A 98% probability that a witness has lied is certainly probable cause to indict under perjury or "false statements" statutes, and needs very little additional evidence to push that to "beyond reasonable doubt". It is also compelling evidence in civil suits under theories such as malicious prosecution.
Are prosecutions for "false statements" or perjury an really an adequate deterrent to defendants lying, especially about the circumstances surrounding police stops, searches, arrests, and incriminating statements?
The threat of such prosecution will be more effective in some kinds of cases than others. In a domestic violence case involving people with no criminal record, it might be decisive. In a case where a defendant faces Lethal Injection it would be nearly irrelevant. In intermediate cases, it combines with de facto deterrents, such as less favorable plea bargains offered by the prosecutor, or more severe sentencing by the court if the defendant is convicted.
Responsible officials in several municipal and state police departments and the U.S. Department of Justice have stated that they are convinced that the threat of prosecution made credible by the Protocol will reduce the volume of false allegations against Police Officers, Corrections Officers, and Border Patrol Agents. This will ease the courts' burden of suppression hearings, and enable law enforcement agencies to keep honest officers at their usual assignments. It will also enable them to focus investigative attention on the remaining complaints, and get rid of rogue officers sooner.
Before implementing a reform as significant as the paired testing, shouldn't courts and legislators wait until the public wants it?
Perjury is rampant. Paired testing offers a way to have a judicial system that places a high value on truth, and in which the incidence and impact of perjury is sharply reduced. Presumably, the judicial system and the public want that.
How can paired testing ever become a reality? No infrastructure even exists to handle certificate issuance, roster maintenance, "blind review" processing, etc., etc.
Major reforms and novel technologies almost invariably require new infrastructure. Many existing non-profit institutions exist that have the integrity and expertise to organize and oversee these functions. Entities to handle the infrastructure such as certificate issuance, roster maintenance, and "blind review" processing, can form quickly either under the auspices of national organizations such as the APA or the AAPP or under the auspices of the courts.
Any number of organizations may undertake the role of disinterested intermediary. A professional journal, experienced in rigorously "blind referee" procedures, would be ideal for the task. A prestigious Law School might undertake it, perhaps through its Law Review. Professional organizations such as the APA or the AAPP might create an entity to do it under their supervision. Some judges may even elect to assume the burden themselves, albeit at the risk of having their caseloads increased.
If the non-profits choose not to offer the service, that would open opportunities to the for-profit sector. One nice feature of our free capitalist economy is that is usually profitable to provide an important new service, and no shortage of entrepreneurs ready to seek those profits.
Forensic DNA technology faced a similar problem a few years ago, when there were no experienced laboratories, no established procedures for them or the courts to follow, no data about the frequency of alleles in the population, etc. One might equally have argued against the automobile on the ground that there were no gas stations.
Isn't it out of place for non-lawyers to be telling the legal profession what to do?
The members of the polygraph community know the most about the production of polygraph results and the people who produce them. Their standing to offer guidance as to their effective use is at least equal to that of any other professional community. One might equally suggest that the molecular biology community was out of place to provide guidance in the optimal forensic use of DNA.
Would the public tolerate a woman who is bringing charges of child molestation or abuse being denied the right to testify about them?
False allegations of child abuse are frequent. If an accused elects to be examined under the Protocol and the accuser refuses, that should establish reasonable doubt as to the guilt of the accused, and probably a strong suspicion of malicious prosecution in the mind of the viewers.
How will paired testing affect the budgets and workloads of public defenders offices?
Public defenders will expedite their workloads by pleading out guilty clients more quickly, releasing many innocent clients quickly via paired testing, and being able to focus their resources on their most deserving clients.
The Protocol cannot possibly address the myriad rules and procedures whereby jurisdictions allocate the costs of investigators and experts pursuant to indigent litigants' needs. Most jurisdictions will adjust their budgets to sensibly re-allocate the savings resulting from reduced and expedited litigation. In those that don't, the mess will be resolved through the political process.
Will enough real-world cases fit within the scope of the Protocol to make a difference?
He said / She said situations are common in many types of litigation in practically every forum in every state.
A - Criminal proceedings: In every state, every day, there are hearings for the suppression of evidence based on alleged Fourth Amendment violations.
B - Civil proceedings: In few civil cases do the parties stipulate as to all the dispositive facts. Many, if not most, are settled based upon the parties' assessment of which side's version of the facts the jury is likely to believes, or decided by jury believing one side's version.
C- Family court: Every day in every state there are hearings to resolve
allegations of abuse or perverse behavior, often incident to divorce and
PARTIAL SCOPE OF APPLICABILITY
Criminal: Date rape - consent or not.
Assault? or self-defense
Confession: Coerced or not
Evidence: Planted or not
Grounds for search and seizure: True as claimed, or not
Incriminating statement to informant: Made as reported or not
Excessive injury claims
Discovery improprieties (e.g. shredding documents)
Concealment of assets
Family: Divorce and Custody:
Alleged domestic violence
Alleged abuse of children
Labor: Alleged racial discrimination
Alleged sexual harassment
Why does the Protocol require that examinations be conducted simultaneously?
The Protocol recommends simultaneous examinations but does not require them. It recommends them in order to prevent an early result possibly contaminating a later one. When simultaneity is impractical for any reason, there are other ways for courts to address the risk. The Protocol makes absolutely no distinction between result of examinations that are conducted simultaneously, and those that are not.
If a witness has previously been examined regarding disputed facts, does he have to undergo another examination about those same facts?
The Protocol requires examinations be conducted by examiners selected from the Roster by the Court, so previous testing doesn't count. Allowing it would open a rat's nest of possible conflict-of-interest, corruption, and undue-influence problems.
Could the Protocol result in the exclusion of one person's testimony because another person refused to undergo examination?
No. There is no combination of circumstances where that would happen. Where the testimony of two or more witnesses from the same side would corroborate each other, and some of them refuse to be examined, the one who refuse are regarded as deceptive, and they are subject to the provisions pertaining to exclusion and impeachment of untrustworthy testimony. The others may undergo examination, and depending on the outcome may testify.
Don't competing experts usually cancel each other out (and waste a lot of time and money?)
Under the Protocol, examiners never testify against each other in court. The Protocol sharply curtails the possible impact of a reviewer with an agenda. Regardless of a reviewer's methodological preference, the report must specify "serious deviations from the norms of good practice of the methodology employed by the original examiner". Two additional reviewers must then agree in order to reverse the original examiners conclusion. If only one agrees, then the effect is only to void, not change, the result.
Does the Protocol permit truthful witnesses to have the polygraphists testify on their behalf?
No. Under the exclusionary option, the truthful witness testifies unopposed as to dispositive facts. That is a tremendous advantage to his side, and sufficient. Under the impeachment option, counsel for the witness's side will almost always elect not to have his witness testify regarding the facts where he was found deceptive, which works out to the same thing. If he chooses to put his witness on the stand he must stipulate to the probability of deception.
Either way precludes any need for the polygraphists to testify, thus saving the court and the litigants having to qualify the polygraphists as experts, and pay them for time in court.
Won't the Protocol mean that more time will be required of expert witnesses, increasing costs?
The protocol applies to facts about which it is highly likely that at least one witness is lying, and about which neither is likely to be mistaken. Experts often have honest differences of opinion. The Protocol will seldom be allowed to challenge expert testimony.
If a witness refuses to undergo examination, who pays the cost for the other witnesses' tests?
The text is clear. The DI side pays both sides' costs. If the results are such that no strong inference can be drawn, the invoking party pays them. A refusal is the same as a DI; therefore the refusing side pays the costs for the other side's examination.
Research has established that eyewitness testimony is often highly unreliable. How does the Protocol accommodate that?
The Protocol excludes from its scope witnesses who may be honestly mistaken. It leaves to the court to decide whether a witness might reasonably be honestly mistaken about who or what he or she saw. The research identifies many factors that affect the reliability of eyewitness testimony - how well the witness knows the person, how long the incident lasted, the lighting conditions, the distance between them, etc. There is an almost infinite gradation between a witness recognizing a stranger who she saw for a few seconds, say, and recognizing her mother.
Many cases have large numbers of witnesses. Doesn't the Protocol threaten to drown the process in a sea of polygraphy?
Not many cases have more than one or just a few dispositive facts, about which witnesses directly contradict each other, and about which neither witness can reasonably be mistaken. The more such pairs there are, the greater the likelihood that paired testing will work to the advantage of the honest litigant.
Some attorneys will no doubt attempt to invoke the Protocol for facts or witnesses that are beyond the explicit limits of its scope. Courts, exercising reasonable discretion at the behest of opposing counsel, will thwart them.
For judges, evidentiary rulings are a minefield of possible reversible error. Doesn't the Protocol add new mines to the minefield?
The Protocol's criteria, for determining whether facts are dispositive and witnesses appropriate, are sufficiently clear that reversals should be few. The possibility of reversal helps assure reasonable discretion in applying them. Trial judges expect to be appealed; it's being reversed that irks them. Close calls will presumably be decided in favor of parties invoking the protocol, because of the risks involved in invoking it for a deceptive witness.
Who pays the cost for polygraphy under the Protocol? And doesn't the review process introduce serious uncertainty as to prospective costs?
The invoking party pays all costs, unless there results a strong inference that the other side's witness was deceptive. If a party requests that a result be reviewed, that party pays the costs of the review unless the original result is reversed. It is true that neither side knows at the outset whether examinations will be reviewed, or witnesses re-examined. Even in the worst case, where after reviews and re-examinations there is no strong inference, the costs of polygraphy will be usually be small in comparison to the total costs of a litigation.
Who pays the cost of reviews?
The side requesting a review does so with the full understanding that it will have to bear the costs if the original opinion is not reversed.
Most Frequent Single Question (composite): HELP! My opponent knows he's lying through his teeth. His strategy is to bankrupt me with motions, hearings, interrogatories and depositions. Is paired testing available in (name of state)?
Not yet. But soon.
Doesn't the Protocol conflict with state laws? For example, states forbid penalizing a refusal to be polygraphed, prohibit authorities to polygraph the complaining witness in sexual assault cases, etc.
The Protocol is generally in accord with most state laws. Where the laws of a particular jurisdiction preclude its application to particular cases or classes witnesses, the courts of that jurisdiction will not apply it to those cases or witnesses.
Rules against penalizing refusal to be polygraphed make sense when applied to single examinations by examiners of unproven proficiency, whose results cannot be quantified and are not admissible evidence. They do not make sense in the context of paired testing.
In all other contexts the rules can remain in full force. Given the odds, rational deceptive litigants will rarely invoke the protocol, and rational honest litigants will not decline when they do. Where necessary, rules can change to accord with good sense. When the interests of efficiency, fairness, and justice can be facilitated by adapting rules to new conditions, people of good will see that they are changed.
Does the Protocol violate Federal Law or the U.S. Constitution?
In active consultation with attorneys, the protocol has been carefully crafted to conform to the Federal Rules of Evidence, which have been adopted nearly verbatim by most states, and to the body of case law interpreting them. It has also been crafted to conform to the requirements of the constitution, particularly in regard to the rights of criminal defendants under the Fourth, Fifth, and Fourteenth Amendments and to the body of case law interpreting them. Indeed, it offers the prospect of a valuable de facto extension of the Confrontation Clause that may significantly help innocent persons accused of crime.
Doesn't the EPPA prohibit employers from compelling employees to be polygraphed?
The Protocol and EPPA will seldom interact, though one can suppose rare hypothetical cases. If an employee fired for alleged wrongdoing sued the employer, EPPA might be construed to bar the employer (though not the employee) from invoking the Protocol.
Don't most jurisdictions forbid introducing polygraph results in court?
Existing statutes and case law affecting the polygraph are either inapplicable to the Protocol, or are grounded in reasoning which makes them favorable to it. Statutes and precedents involving the polygraph that are reasonably applied to results of single examinations conducted by examiners of unproven proficiency and offered as scientific evidence, do not reasonably extend to results under the Protocol.
The statutes and precedents restrict:
· The introduction of polygraph results
· By examiners of unproven proficiency
· Of examinations of individual witnesses
· As scientific evidence
Because they understandably find such evidence insufficiently trustworthy.
The Marin Protocol:
· Bars testimony found to be dubious
· As a combinatorial result of examinations of multiple witnesses
· Conducted by examiners of proven proficiency
Because such evidence is insufficiently trustworthy.
The Protocol does not dictate its own adoption by any jurisdiction. It recognizes and covers many subtle issues that must be addressed if paired testing is to be utilized effectively. It offers jurisdictions that choose to utilize paired testing a guide to doing it right.
How does the SCHEFFER decision affect the prospect of the Protocol being implemented?
* The Supreme Court in Scheffer held that it was not unconstitutional for jurisdictions to exclude the polygraph, per se, as scientific evidence. The opinion quoted by the voter represented the views of only four of the nine members of the court. Four of the Justices refused to concur in the quoted portion of the Opinion, pointing out that the doctrine of the jury as the lie detector had been diminished through time even as to the ultimate facts of a case. They pointed out that the last remaining reed of the doctrine is Fed. Rule Evid. 704(b): "No expert witness testifying with respect to the mental state or condition of a defendant in a criminal case may state an opinion or inference as to whether the defendant did or did not have the mental state or condition constituting an element of the crime charged or of a defense thereto. Such ultimate issues are matters for the trier of fact alone." One Justice dissented altogether.
The opinion also stated: "Although the degree of reliability of polygraph evidence may depend upon a variety of identifiable factors, there is simply no way to know in a particular case whether a polygraph examiner's conclusion is accurate, because certain doubts and uncertainties plague even the best polygraph exams. Individual jurisdictions therefore may reasonably reach differing conclusions as to whether polygraph evidence should be admitted."
That point, fairly made regarding single examinations by examiners of unproven proficiency, works in favor of the Protocol, which provides "a way to know in a particular case" that the conclusions drawn from multiple polygraph examinations are, to a very high degree of probability, accurate.
In the portion of the opinion concurred in by eight Justices, the Court acknowledged the exclusion of unreliable evidence, writing: "State and federal governments unquestionably have a legitimate interest in ensuring that reliable evidence is presented to the trier of fact in a criminal trial. Indeed, the exclusion of unreliable evidence is a principal objective of many evidentiary rules. See, e.g., Fed. Rule Evid. 702; Fed. Rule Evid. 802; Fed. Rule Evid. 901 . . ."
The Protocol is novel - there really is no precedent. The rules cited above limit admissibility of hearsay, expert testimony, and documents. The Federal rule that most closely applies to the Protocol is Rule 403, which reads: "Although relevant, evidence may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury, or by considerations of undue delay, waste of time, or needless presentation of cumulative evidence."
That rule has been adopted, more or less verbatim, in most states. While it does not compel courts' to exclude testimony in accordance with the Protocol's provisions, it certainly permits them to do it. Rarely if ever will a court be reversed for a determination that testimony with a 98% likelihood of being perjured carries the danger of being "misleading to the jury".
Didn't Chambers v Mississippi holt it unconstitutional for a judge to prevent a criminal defendant's witnesses from testifying?
Chambers v Mississippi acknowledged judges' reasonable discretion to exclude misleading evidence, then found the trial judge's exclusion of three reputable witnesses to whom a person other than the defendant had confessed the crime to be unreasonable.
The Court has held that the exclusion of testimony for myriad reasons does not violate due process, among them being testimony found prejudicial, irrelevant, non-probative, inflammatory, cumulative, time-wasting, hearsay, and of course, misleading. In a recent New York murder case the defendant claimed that the president of the United States had ordered the crime, and tried to get him subpoenaed as a witness. It didn't work, of course, but nice try.
Supposing that precedents limiting the admissibility as scientific evidence of single-test results conducted by uncertain examiners do not apply to paired testing. How can you predict what courts will do?
Obviously there are no case precedents for a brand new Protocol. How could there be? But the underlying principle is expressed by the Rules of Evidence of practically every state, and goes back to the English Common Law. It is the principle that underlies the exclusion of hearsay testimony.
EXAMINERS AND CERTIFICATION
Why the elaborate Litigation Certificate Process? Should it not suffice that an examiner has graduated a good school, has substantial real-world experience, and has good reputation within the profession?
The high confidence warranted by the strong inferences that are derived from examinations pursuant to the protocol requires the proven individual proficiency of the examiners. General credentials are no substitute. Neither is standing in the profession, given that the NAS found about half of examiners achieve accuracy of less than 86%, and about a quarter less than 81%.
What guarantee is there that examiners will perform as well in real cases as they did during their certification?
Some examiners might not perform as well in live cases as they did during their certification. That is why there is a review procedure.
Why not just establish a Pass-Fail system for certification? Won't publishing examiners' accuracy and conclusive rates allow lawyers to cloud the issues?
The strength of the Protocol lies in the high level of confidence its strong inferences warrant when they are generated. It reasonably recommends minimum levels accordingly. Published ratings give courts the opportunity to select the best available examiners. The Protocol recognizes, however, that if circumstances such as a sparse selection of high-scoring certificate holders require a state or court to either relax its criteria or forego the Protocol altogether, then it is better to relax the criteria.
The certification process will involve significant cost to the applicants. Under a pass-fail system, a near-miss means the total loss of the investment. That is likely to discourage many capable applicants, especially in the first few years when they will be urgently needed.
For the first several years, the demand for examiners may be intense, while the availability of people certified above a high threshold would be limited. The protocol should not become a source of delays, while litigants wait on examiners' calendars. The published proficiency approach permits courts and litigants to choose to use examiners whose proficiency may be a little less than ideal, in order to move cases along.
Numeric rating is fully open about the tested proficiency of the examiners in the roster. What you see is what you get. If a pass-fail system made the performance numbers of passing applicants available, it would be no different. If it hid them, that would provide lawyers with opportunities for challenge and obfuscation.
There are several schools of thought within the polygraph profession, each with its icon and followers. Shouldn't paired testing wait until the profession reaches a consensus on the best approach?
The 'icons" took their stands years ago. They've stood their ground ever since and their followers may still be standing there long after the icons have died or retired. Data collected in the course of certifying examiners may help resolve the issues.
It may not be desirable for everyone to march to the same drum. As new and better "drums" are developed, both the Protocol and the profession should be flexible enough to accommodate them. Also, it may be that no one method is right for everyone. Methods that work well for some examiners may not work as well for others, so that examiners need to be able to choose from among several alternatives the methods best suited to their individual strengths and weaknesses and personalities.
It may emerge that practitioners employing some methods show significantly higher (or lower) overall proficiency than those employing others. It may emerge that each method has its more and its less proficient examiners, but that no method is consistently superior to the others. Either result should help resolve the interminable and often unseemly wrangling within the profession. The "icons" and their disciples will have the same opportunity as everyone else to show what they can do, and will have to live with the results.
Won't the certification and review of examiners and examinations be something of a crapshoot, depending on whether the preferences of certifiers and reviewers differ from those the people whose work they are evaluating?
Paired testing is neutral with respect to competing techniques and methodologies. It certifies examiners based on only one criterion - their personally demonstrated proficiency. Examiners are certified based on their proficiency with whatever methodology they use. When reviewers challenge an original examination, they must document significant deviations from the standard of whichever method the original examiner employed.
Why does the Protocol permit examiners to opt out of the roster and keep their scores secret?
The Protocol is only concerned with examiners who are or aspire to be on the roster of persons eligible to conduct examinations pursuant to it. If an examiner chooses not to be on the roster, and wishes to keep his scores secret, that is his decision. It is no more the business of the Protocol to weed marginal practitioners from the PDD profession, than it is to establish an elite.
Why doesn't the Protocol permit an examiner who has been over-ridden on review to come into court to defend his work product?
Cases are about the claims and interests of the parties. The examiner is not one of the parties.
(C) 2001 by Jonathan Marin
Reasons for Favoring Adoption of the Proposed ASTM Standard for Paired Testing
Expanded Version of original article published in Polygraph, presenting the paired test concept
UNITED STATES vs. SCHEFFER, AMICUS CURIAE BRIEF IN SUPPORT OF SCHEFFER COMMITTEE OF CONCERNED SOCIAL SCIENTISTS (Charles F. Peterson, Counsel of Record and Charles R. Honts, Ph. D., Primary Author)
UNITED STATES vs. SCHEFFER, AMICUS CURIAE BRIEF IN SUPPORT OF THE GOVERNMENT CRIMINAL JUSTICE LEGAL FOUNDATION