TY - GEN
T1 - Die Chancen von KI bei der Personalauswahl nutzen ̶ Chancen für die Förderung von Chancengerechtigkeit und Diversität
T2 - HR Performance Special, 4/2022, 12–14
AU - Röhner, J.
AU - Schütz, M.
AU - Schütz, Astrid
N1 - J. Röhner and M. Schütz share first authorship
PY - 2022/12
Y1 - 2022/12
N2 - Background: Medical expert witness assessments (MEWAs) evaluate case validity on the basis of both psychometric and non-psychometric modalities. Although symptom validity tests (SVTs) and performance validity tests (PVTs) are widely used, many thresholds were developed in analogue and known-groups validation contexts, and their transferability to MEWAs remains uncertain. Prior research validated a Criteria-Based Validity Assessment (CVA) as a multimodal framework for assessing case plausibility, but it remained unclear whether CVA and psychometric thresholds perform similarly across different legal contexts, such as pension insurance versus accident insurance evaluations. Objective: This study aimed to investigate differences in CVA, SVT, and PVT results across pension and accident insurance cases. Methods: A total of 721 MEWAs (572 pension; 149 accident) were analyzed. CVA criteria were rated by trained raters, and response biases were assessed psychometrically using the SIMS and the ASTM. Two-component beta-binomial mixture models were applied to derive CVA plausibility thresholds. Results: Mixture modeling reliably identified a distinct bimodal distribution of conspicuous CVA criteria counts, interpreted as plausible and implausible subgroups in both pension and accident cases, with ≥4 conspicuous CVA criteria representing the optimal plausibility threshold. Pension claimants showed significantly higher SIMS scores and lower ASTM scores than accident claimants, independent of case plausibility. Conclusion: CVA can be used to assess validity information collected from multiple data sources (longitudinal and cross-sectional) and to assess the validity of health-related claims across different legal settings. CVA, however, was not a tool for measuring context-specific response biases. Cases with ≤2 conspicuous CVA criteria may be valid, cases with three conspicuous criteria require individual examination because feigning is possible but not certain, and ≥4 conspicuous criteria supported a conservative classification of CVA-defined case implausibility.
AB - Background: Medical expert witness assessments (MEWAs) evaluate case validity on the basis of both psychometric and non-psychometric modalities. Although symptom validity tests (SVTs) and performance validity tests (PVTs) are widely used, many thresholds were developed in analogue and known-groups validation contexts, and their transferability to MEWAs remains uncertain. Prior research validated a Criteria-Based Validity Assessment (CVA) as a multimodal framework for assessing case plausibility, but it remained unclear whether CVA and psychometric thresholds perform similarly across different legal contexts, such as pension insurance versus accident insurance evaluations. Objective: This study aimed to investigate differences in CVA, SVT, and PVT results across pension and accident insurance cases. Methods: A total of 721 MEWAs (572 pension; 149 accident) were analyzed. CVA criteria were rated by trained raters, and response biases were assessed psychometrically using the SIMS and the ASTM. Two-component beta-binomial mixture models were applied to derive CVA plausibility thresholds. Results: Mixture modeling reliably identified a distinct bimodal distribution of conspicuous CVA criteria counts, interpreted as plausible and implausible subgroups in both pension and accident cases, with ≥4 conspicuous CVA criteria representing the optimal plausibility threshold. Pension claimants showed significantly higher SIMS scores and lower ASTM scores than accident claimants, independent of case plausibility. Conclusion: CVA can be used to assess validity information collected from multiple data sources (longitudinal and cross-sectional) and to assess the validity of health-related claims across different legal settings. CVA, however, was not a tool for measuring context-specific response biases. Cases with ≤2 conspicuous CVA criteria may be valid, cases with three conspicuous criteria require individual examination because feigning is possible but not certain, and ≥4 conspicuous criteria supported a conservative classification of CVA-defined case implausibility.
KW - Accident insurance
KW - Criteria-Based Validity Assessment
KW - Medical expert witness assessment
KW - Pension insurance
KW - Performance validity test
KW - Symptom validity test
M3 - Andere Veröffentlichung (nicht peer-reviewed)
T3 - HR Performance
ER -