Advancing selection models for psychological fitness assessment in highly complex occupational roles

Positive response bias (PRB) represents one of the most significant threats to the validity of psychological assessment in high-stakes decision-making contexts, including aviation personnel selection. The present thesis addresses PRB through three interrelated studies aimed at documenting its structure in an ecological sample of pilot candidates, developing a multidimensional self-report measure, and evaluating the utility of implicit behavioural measures for its detection. Study 1 examined MMPI-2-RF profiles of 198 civil pilot candidates assessed at the CAVOK centre in Lonate Pozzolo during the final phase of selection. Both validity scales were significantly elevated relative to normative means, with K-r showing the larger effect (d = 1.01 vs. d = .70 for L-r). Cluster analysis identified three subgroups: a sophisticated K-dominant profile (47%), a global underreporting profile combining high L-r and K-r (29%), and a low-defensiveness profile (24%), highlighting marked heterogeneity in the expression of PRB. Study 2 described the development and preliminary validation of the Response Bias Inventory (RBI), a 25-item questionnaire organised into five subscales (Anger Denial, Superiority, Aggressiveness Denial, Assumption of Responsibility, Social Irreproachability). In an experimental sample of 642 participants (honest vs. instructed-faking condition), the RBI showed good internal consistency at the total-score level (α = .85) and acceptable discriminative accuracy (AUC = .72). Study 3 compared 43 students under honest instructions, 43 students under faking instructions, and 28 genuine pilot candidates, administering the RBI alongside the MMPI-3 L and K validity scales while simultaneously recording mouse-tracking and webcam-based eye-tracking data. Pilot candidates and faking students obtained statistically equivalent scores on the RBI, Scale L, and Scale K, with pilots numerically exceeding the faking group on Scale K. Behavioural measures based on cursor trajectory did not differentiate the three groups; only movement onset latency distinguished pilots from students, emerging as a population-level rather than faking-specific marker. The findings support an assessment approach that combines multidimensional self-report instrumentation with a typological interpretation of validity profiles, while highlighting the current limitations of implicit behavioural measures for detecting K-dominant PRB.