Zur Testfairness der aussagepsychologischen Begutachtungsmethodik: eine Bayes-statistische Analyse anhand von 2000 Gutachten

Zusammenfassung
In aussagepsychologischen Begutachtungen wird geprüft, ob einer Aussage ein Erlebnisbezug unterstellt werden kann. Eine Beauftragung erfolgt meist, wenn die aussagende Person Merkmale aufweist, deren Einordnung psychologische Expertise verlangt. Darunter fallen ein junges Alter, eine Intelligenzminderung oder psychische Störungen. Bislang ist jedoch wenig darüber bekannt, ob und wie sich diese Merkmale auf das Begutachtungsergebnis auswirken. Zudem gibt es Kritik, dass die Begutachtungsmethodik bei diesen Personengruppen zu Benachteiligungen führt oder nicht anwendbar sein könnte. Um diese Kritik der mangelnden Testfairness zu untersuchen, wurde in der vorliegenden Studie ein Datensatz von 2005 Gutachten mittels Bayesstatistischer Regressionsanalysen ausgewertet. Die Ergebnisse zeigen, dass weder demografisch-persönliche noch klinisch-psychologische Merkmale der aussagenden Person einen klaren Zusammenhang mit einem bestimmten Begutachtungsergebnis aufweisen. Die 8 berechneten Regressionsmodelle weisen vielmehr unsichere Effektschätzungen auf, was nahelegt, dass andere Merkmale für das Begutachtungsergebnis bedeutsamer sein müssen. Eine ergänzende korrelative Nachanalyse zwischen kritikrelevanten Merkmalen der aussagenden Person und den Verhinderungsgründen für eine Erlebnisannahme ergab ebenfalls keine Hinweise auf eine (systematische) Benachteiligung bestimmter Personengruppen. Insgesamt sprechen die Befunde für eine ausreichende Testfairness der aussagepsychologischen Begutachtungsmethodik. Eine eingeschränkte Anwendbarkeit der Methodik bei Personen jungen Alters, mit einer Intelligenzminderung oder Symptomen einer psychischen Störung lässt sich nach aktueller Befundlage nicht annehmen.

Bayesian statistical regression analysis of 2,005 forensic expert reports examining relationships between witness characteristics (age, intellectual disability, mental disorder symptoms) and assessment outcomes, supplemented by correlational analysis of reasons for rejecting genuine experience attribution.

Background & Context

Statement validity assessment (SVA), known in German forensic psychology as "aussagepsychologische Begutachtung," represents a cornerstone methodology for evaluating the credibility of witness testimony in legal proceedings. These assessments determine whether a statement can be attributed to an actual experience, and are typically commissioned when testifying individuals exhibit characteristics requiring psychological expertise—such as young age, intellectual disabilities, or mental disorders.

Despite widespread use in German-speaking criminal justice systems, a critical gap has existed in empirical research: little is known about how these witness characteristics influence assessment outcomes. Concerns have been raised that the methodology may disadvantage certain groups or may not be fully applicable to them, raising fundamental questions about test fairness—whether the assessment process systematically produces biased outcomes based on demographic or clinical characteristics rather than the actual truthfulness of statements.

This 2026 study by Jäckel, Dege, and Posten represents the largest empirical investigation of test fairness in statement validity assessment to date, addressing longstanding criticisms through modern Bayesian statistical analysis of over 2,000 real-world expert witness reports.

Research Design & Methodology

The study analyzed a dataset of 2,005 expert witness reports using Bayesian statistical regression analysis. This exceptionally large sample drawn from actual forensic practice provides unprecedented statistical power to detect potential bias effects. The researchers employed Bayesian regression rather than traditional frequentist approaches, allowing for more nuanced quantification of uncertainty in effect estimates.

The analysis examined relationships between assessment outcomes and two categories of witness characteristics:

• Demographic-personal variables: Age of the witness, particularly focusing on young witnesses who comprise a substantial proportion of credibility assessments
• Clinical-psychological variables: Presence of intellectual disability (cognitive impairment) and symptoms of mental disorders
• Outcome measures: Whether the expert concluded the statement could be attributed to genuine experience versus other outcomes (not credible, indeterminate)

Eight regression models were calculated to examine different relationships between predictor variables and assessment outcomes. The researchers also conducted supplementary correlational follow-up analyses between criticism-relevant characteristics and the reasons that prevented assuming genuine experience—essentially examining whether certain groups were disproportionately rejected for specific methodological reasons.

Results & Key Findings

The study's central finding contradicts fears of systematic bias: neither demographic personal nor clinical psychological characteristics of testifying individuals exhibit a clear association with assessment outcomes. This represents strong evidence for test fairness across vulnerable populations.

Key statistical findings included:

• The eight calculated regression models displayed uncertain effect estimates, indicating weak or negligible relationships between witness characteristics and outcomes
• The uncertain effect estimates suggest that other characteristics must be more significant for assessment outcomes—namely, the actual content quality and consistency of the statements themselves
• The supplementary correlational analysis between criticism-relevant characteristics and reasons preventing experience attribution revealed no evidence of (systematic) disadvantaging of specific population groups

These null findings are methodologically significant precisely because of the large sample size and sensitive Bayesian methods employed. The absence of detectable bias effects with this level of statistical power provides robust evidence that the methodology functions fairly across diverse witness populations.

Discussion & Significance

This research addresses a critical gap in forensic psychological methodology by providing empirical evidence regarding test fairness concerns. The findings support the theoretical foundation of statement validity assessment—that credibility judgments should be based on statement characteristics (logical consistency, richness of detail, complexity) rather than witness demographics or clinical status.

The results have important implications for ongoing debates in German forensic psychology. Critics have argued that SVA methodology may be inappropriate or systematically unfair when applied to young children, individuals with cognitive impairments, or those with mental health conditions. Overall, the findings speak to adequate test fairness of statement psychological assessment methodology. A restricted applicability of the methodology for persons of young age, with intellectual disability, or symptoms of mental disorder cannot be assumed based on current evidence.

The Bayesian approach provides advantages over traditional significance testing by quantifying uncertainty in effect estimates rather than simply rejecting null hypotheses. The finding of uncertain (rather than merely "non-significant") effects suggests that if bias exists, it is too small to be practically meaningful even with 2,005 cases analyzed.

Limitations & Considerations

While the sample size is exceptional, several methodological considerations warrant attention. The study analyzed existing expert reports rather than conducting experimental manipulations, meaning it reflects real-world forensic practice but cannot control for all confounding variables. The quality and thoroughness of individual assessments may have varied across the 2,005 reports included.

The analysis focused on outcomes (final conclusions) rather than examining the intermediate assessment processes. It remains theoretically possible that evaluators apply different standards during analysis but reach similar overall conclusions across groups. Additionally, the study examined group-level patterns; individual cases of inappropriate bias cannot be ruled out even if systematic patterns are absent.

The research was conducted within the German forensic psychology system. Generalizability to other legal and cultural contexts where SVA is applied differently may be limited.

Practical Applications

For forensic examiners, these findings provide empirical support for applying standard SVA methodology across diverse witness populations without categorical exclusions based on age, cognitive ability, or mental health status. This does not eliminate the need for individualized assessment—examiners must still carefully evaluate each witness's specific capabilities and any factors affecting statement quality—but it suggests the methodology itself does not introduce systematic bias.

For legal professionals and courts relying on credibility assessments, the research offers reassurance that properly conducted SVA evaluations can provide fair assessments regardless of witness characteristics. This is particularly important given that witnesses with these very characteristics (young age, intellectual disability, mental health symptoms) are precisely those for whom expert psychological assessment is most frequently requested. The findings support continued use of the methodology while highlighting that assessment quality depends on the actual content and consistency of statements rather than witness demographics.