Home | Services to Attorneys | Vitae | Ask the Doc

False Confessions: Annotated Clinical Research

Joe Wheeler Dixon, PhD, JD

The following annotated research findings have to do with false confessions and police interrogation techniques. This listing was prepared for a defense attorney who was representing a adolescent defendant with borderline intelligence who had arguably falsely confessed to a crime.
Feel free to utilize the references below, but understand that research in this field is constantly changing, and the references below are not up to date.

Prepared for the Hon. William T. Beauregard, Esq.
Joe W. Dixon, PhD, JD

    The following synopses can be used to generate questions for direct examination or re-direct of the expert forensic psychologist you have retained.  Also, this information may be used as a database to develop questions for examination of the opposing counsel’s expert. Most jurisdictions adhere to the doctrine that a false confession will be judicially decided based upon analysis and consideration of the totality of the circumstances, which means police conduct, defendant's conduct, and other circumstances.

    All psychological tests and procedures mentioned below meet or exceed both Frye and Daubert (FRE 2001) admissibility standards. A detailed list of references mentioned below is available; this list is not all inclusive.

Gudjonsson Suggestibility Scale (GSS) instrument used to assess possibility of a false confession

  1. The GSS2 has been factor analyzed and two distinct factors emerge: Yield and Shift, with a coefficient alpha of 0.87 for Yield and 0.79 for Shift, the two factors. Statistically, the scale is functioning as it was designed and is validly measuring the construct it was designed to measure with good internal consistency. The test also has good face validity. Singh & Gudjonsson, 1987; Gudjonsson, 1987; Gudjonsson, 1992; Merckelbach et al, 1998.
  2. Inter-rater reliability of GSS2 is good with correlation coefficients of 0.99 for Yield, 0.99 for Shift, and 0.99 for Total (p < 0.001). Clare et al, 1994. There are several studies on this aspect of test construction with data reported here only for Clare. All studies produced high coefficients indicating different examiners across different subjects obtain consistent results. The instrument is stable and reliable. Singh & Gudjonsson, 1987; Richardson & Smith, 1993.
  3. The SEm is 4.8, that is, I am 95% confident that his true Total suggestibility score on the GSS (Yield + Shift) is 12 +/- 4.8. This is a measure of the reliability of the subject’s score, and gives us an idea of the error associated with the suggestibility score, i.e., I would be wrong only 5% of the time.
  4. Test-retest reliability coefficients range from 0.82 to 0.92 (p < .001) between the two forms of the instrument (GSS1 and GSS2). This is another measure of the instrument’s reliability over subjects, examiners, and time. This data also goes to the issue of validity. GSS1 and GSS2 are so statistically similar they can be used interchangeably as true alternate forms.

Suggestibility, coercion, and confessions - research on juveniles and youthful offenders

  1. Suggestibility has been shown to correlate with several cognitive variables. There is a pronounced and stable inverse relationship with intelligence and with memory. Gudjonsson, 1983, 1988, 1990; Clare & Rutter, 1995; Gudjonsson & Clare, 1995; Singh & Gudjonsson, 1992; Danielsdottir et al, 1993. This increases their susceptibility to giving false confessions. Richardson & Kelly, 1995.
  2. Subjects with low intelligence are much more susceptible to leading questions, confabulate more, and are more acquiescent with interrogators. Clare & Gudjonsson, 1993. This is mediated by prior convictions. Sharrock & Gudjonsson, 1993.
  3. Poor assertiveness, evaluative anxiety, state anxiety, and avoidance coping strategies were significantly correlated with suggestibility scores. State anxiety was highly positively correlated with the Shift score, indicative that the more nervous a subject is in the police interrogations setting, the more likely he can be persuaded with pressure to change his mind and recollections. Gudjonsson, 1988. Generally, the higher the anxiety the more suggestible. Wolfradt, 1998.
  4. Subjects less than 12 years old are significantly more suggestible and are more easily influenced by negative feedback during interrogation (Shift). From age 12 to 16 subjects perform similar to adults on both memory and Yield, but score higher on Shift. After about age 16, there is no difference between normal adolescents and normal adults in these measures. Six studies in this area, the most representative being Gudjonsson & Lister, 1984.
  5. Suggestibility positively correlated with low intelligence, poor memory recall, neuroticism, and social desirability. Gudjonsson, 1983.
  6. Warning normal adults of misleading or tricky questions can significantly reduce their suggestibility to leading questions. Warren, 1991.
  7. Sleep deprivation significantly increases suggestibility, but subjects maintained confidence in their former responses resulting in high suggestibility to leading questions but resistance to coercion to change their answers, e.g., high Yield and low Shift scores. Additional measures of cognitive intactness showed declines from pre-deprivation levels. This group is most akin to low functioning intellectual subjects in overall cognitive ability. Increased suggestibility is a result of induced cognitive deficits. There are three studies on this issue, with Blagrove, 1996, being the most extensive study.
  8. No relationship between suggestibility and mental patients with hallucinations, reality monitoring, or depression have been established. Four studies, with Smith & Gudjonsson, 1995, the most comprehensive. MacFarland & Morris, 1998, found a positive correlation for dysphoric college students.
  9. The more suggestible subjects are, the less accurate they are in recalling details, and the more erroneous information is produced during interrogation. Tully & Cahill, 1984.
  10. False confessors are more suggestible than subjects who actively resist persuasive techniques. Gudjonsson, 1991.
  11. With male forensic subjects, highly suggestible individuals were less likely to change answers with mild negative feedback (high Yield and low Shift). Richardson et al, 1998.
  12. Police interrogation tactics often include deceptive and deceitful practices. Ironically, these tactics are influential in producing false confessions, although police use them to coerce what they believe are guilty subjects to confess. Underwager & Wakefield, 1992. (Review article).
  13. Low IQ subjects are more likely to believe that falsely confessing will have little or no consequences, because of their knowledge of the truth of the matter and belief that truth always wins out. This naiveté renders low functioning subjects at higher risk for falsely confessing than normal functioning subjects. Combined with susceptibility to acquiescence, suggestibility, compliance with authority, and proclivity to confabulate puts low IQ subjects at significantly higher risk for false confession in context of a police interrogation. Clare & Gudjonsson, 1995.
  14. Police officer’s abrupt demeanor resulted in higher Shift scores without much effect on Yield scores. This suggests the two independent measures of suggestibility are differentially affected: Yield by psychological factors and Shift by social factors, as hypothesized. Bain & Baxter, 2000. The greater the social distance between subject and interrogator, the more pronounced the effect. Baxter & Boon, 2000.
  15. Police and authorities do not believe claimed amnesia, depression, and presenting as suggestible and eager to please render a subject unfit for interrogation. In contrast, police do believe confusion and disorientation, heroin withdrawal, paranoid beliefs, and incomprehension of simple questions render a subject unfit. Gudjonsson et al, 2000.
  16. Police tactics that influence outcomes of interrogations include six significant factors (22 factors identified): intimidation, robust challenge, manipulation, questioning style, appeal, and soft challenge. The higher the frequency of these tactics, the more likely courts rule the evidence inadmissible. Pearse et al, 1999.
  17. It is well established that alcohol ingestion alters affective state, principally by lowering anxiety. As hypothesized, alcohol ingestion lowered suggestibility scores with no effect on Shift. Santtila et al, 1999.
  18. Factors in confessing include: younger age or abuse of unlawful substances within 24 hours more likely, and less likely if attorney present or prior convictions with time served. Pearse et al, 1998. No difference in answering questions found with or without attorney, but significantly less likely to waive right to silence or confess with attorney present. Pearse & Gudjonsson, 1997.
  19. Consistent with previous studies, interrogators are generally unable to distinguish truthful from deceptive subjects; additionally, interrogators with special training were poorer at discriminating, although they offered more logical reasons and were more confident in their judgments. Kassin & Fong, 1999.
  20. Right to silence was claimed as understood by 89% of subjects, yet upon close examination, only 11% had full understanding. Level of understanding was linked to intelligence and was not mediated by demographic or criminal justice influences or experiences. Cooke & Philip, 1998.
  21. Ninety-one experienced detectives assessed for discriminate ability to identify truthful from lying suspects; success rate was only 49% - less than chance. Also found lying associated with fewer body movements (hands, feet, arms, & legs), contrary to majority belief of detectives that liars were more active while lying than truth tellers. Vrij & Frans, 1993.

Home | Services to Attorneys | Vitae | Ask the Doc