Measuring Psychological Defense: Reliability, Convergent Validity, and Evidence for Causal Mechanisms—A Scoping Review

Background: Psychological defenses are central to psychodynamic theory, yet their measurement spans distinct observational, projective, self-report, and computational traditions. Whether these methods identify comparable constructs—and whether reliably classified defense patterns constitute distinct causal mechanisms—remains uncertain. Objective: To map the measurement of psychological defense across levels of aggregation; evaluate evidence on inter-rater reliability and convergence between methods; and characterize experimental, formal-modeling, and multichannel approaches to mechanistic validation. Methods: This scoping review followed JBI guidance and was reported with reference to PRISMA-ScR. OpenAlex and PubMed were searched on October 4, 2026, supplemented by one-step forward and backward citation searching. Of 7,902 records identified, 6,375 unique records were screened and 348 were included in the evidence map; 112 full texts were examined. A second assessor independently screened a random sample of 300 records (Cohen κ = 0.63; 95% CI, 0.42–0.79). Descriptive summaries were based on 78 eligible human–human agreement coefficients from 19 sources, selected from a broader chart of reliability estimates. Results: Inter-rater agreement varied across measurement units, instruments, materials, and procedures. The reported median agreement was 0.78 for individual defenses measured from standardized projective or narrative stimuli (23 coefficients), 0.43 for sets of individual defenses predominantly coded from interviews or speech (12 coefficients), 0.70 for defense levels (22 coefficients), 0.74 for styles or summary scales (6 coefficients), and 0.75 for overall defensive functioning (15 coefficients). These descriptive medians are not directly comparable because coefficient types and study conditions differed. Individual Defense Style Questionnaire scales showed variable internal consistency (median Cronbach α = 0.46 across 21 defenses in one sample). Evidence of convergence between independent methods at the individual-defense level was limited. The map identified 22 experimental records and 16 multichannel studies, but no study meeting the review’s proposed criterion for independent, episode-level confirmation across speech, subjective experience, and physiology. Conclusions: Broad indicators of defensive functioning are measured with reasonable reproducibility in several research settings, whereas evidence for consistently identifying particular defenses in naturalistic material is more dependent on measurement design. Current findings cannot determine whether this reflects limitations of instruments, context-dependent expression, or the construct structure itself. Progress requires within-sample comparisons across measurement units, independent-method validation, and preregistered experimental tests of specific causal predictions.