Valor científico de la prueba de credibilidad del testimonio CBCA-SVA.

Accuracy of SVA Assessments

Aquí tenéis un documento judicial para argumentar la chapuza de técnica.

Testing the accuracy of SVA assessments and other lie detection tools may sound straight­forward, but, in fact, is difficult. Accuracy can easily be determined in controlled laboratory based experiments. In such experiments, participants (typically college students) are asked to tell a truth or lie for the sake of the experiment, and lie detectors attempt to distinguish between these truths and lies by using their lie detection tools. The main limitation of Ihis procedure is the lack of ecological validity. What does the ability of an SVA expert to distinguish between lies and truths told by college students in controlled settings say about his/her ability to accurately determine the veracity of a child's statement in an alleged sexual abuse case? Very little according to many SVA experts, because the situations are very different. The same applies to polygraph tests. Being able to detect lies and truths with a polygraph test in a laboratory experiment may well be completely different from being able to delect lies and truths told by suspects in criminal cases. Psychologists therefore agree that in order to evaluate the accuracy of lie detection tools, field studies need to be carried out in which assessments are examined that were made in real life cases.

The difficulty in field studies is determining the ground truth, that is, to establish the truth/innocence status of the examinee beyond doubt. SVA assessments take place in sexual abuse cases. It is often difficult to determine the facts of a sexual abuse case, since often there is no medical or physical evidence. Frequently [he alleged victim and the defendant give contradictory testimonies and often there are no independent witnesses to give an objective version of events. Therefore, in SVA field studies confessions are often used as a criterion. This is problematic, as confessions are not independent from SVA veracity judgements. For example, if the only evidence against the guilty defendant is the incriminating statement of the child, which is often the situation in sexual abuse cases, it is unlikely that the perpetrator will confess to the crime if the incriminating statement is of poor quality. On the other hand, if a false incriminating statement is persuasive and judged to be truthful by a SVA expert, the chances for the innocent defendant obtaining an acquittal decrease dramatically (at least in Germany), and it thus may be beneficial to plead guilty in order to obtain a reduced penalty (Steller and Kohnken, 1989). In summary, poor quality (e.g. unconvincing) statements decrease the likelihood of obtaining a confession and high quality (e.g. convincing) statements increase the likelihood of obtaining a confession, regardless of whether a statement is truthful or fabricated.

The same reasoning applies to polygraph tests, which are typically conducted when the veracity of the suspect's statement cannot be determined in other ways. Suspects who are found guilty in the test may well believe that they cannot demonstrate their innocence anymore, which may result in a false confession. On the other hand, guilty examinees who pass the polygraph test have no reason to confess.

The interdependence between SVA and polygraph outcomes and confessions may result in inflated accuracy figures in field studies. Incorrect assumptions of guilt that subsequently lead to false confessions will be classified as 'hits' according to the 'confession equals guilt' rule. No confessions from guilty examinees who pass the test will also be classified as 'hits' when the 'no confession of the examinee equals innocence' rule is used. However, most studies use a somewhat different rule regarding innocence. An examinee is only classified as innocent if someone else confesses to the crime. This rule still causes trouble. Many cases where the guilty examinee passes the test will remain unsolved, because it is unlikely that an innocent person will confess in this case. Cases without confessions will not be included in the field study because of a lack of ground truth. Consequently, the mistake made by the examiner will remain unnoticed.

A known error rate in CBCA field research therefore does not exist. Accuracy rates of CBCA field studies have been calculated in two CBCA field studies (Esplin. Boychuk and Raskin, 1988; Parker and Brown, 20(H)) but both studies were Hawed. Amongst other obstacles, there were problems with how the ground truth was established Vrij (2005. in press). Error rates regarding the Validity Checklist and the SVA method as a whole only have been reported in the flawed Parker and Brown (2000) study Vrij (2005, in press). Laboratory CBCA research has revealed that, on average, 73 % of the truths and 72 % of the lies were correctly classified, resulting in a known error rate of just below 30 % for both truths and lies Vrij (2005, in press). However, the vast majority of experimental studies used undergraduate students as participants. The few studies with children as participants revealed similar accuracy rates.

SVA Assessments and the Daubert Test

In Daubert v. Merrel Dow Pharmaceuticals, Inc. (1993), the United States Supreme Court set out guidelines for admitting expert scientific evidence in the federal (American) courts. The following guidelines are provided by the Supreme Court (Honts, 1994): (1) Is the scientific hypothesis testable?, (2) Has the proposition been tested?. (3) Is there a known error rate?. (4) Has the hypothesis and/or technique been subjected to peer review and publication? and (5) Is the theory upon which the hypothesis and/or technique based generally accepted in the appropriate scientific community? Table 5.1 summarizes my answers to these questions for SVA assessments.

The prediction that truthful statements will obtain higher CBCA scores than false state­ments can be tested in scientific research, although this is not an easy task in field research given the problems with establishing the ground truth. The answer to the first Daubert question is therefore 'yes' for CBCA laboratory research but 'problematic' for CBCA in field studies. Some underlying assumptions of the Validity Checklist are also difficult to test