ADI™-R

To administer the ADI–R, an experienced clinical interviewer questions a parent or caretaker who is familiar with the developmental history and current behavior of the individual being evaluated, and uses standardized procedures to code the responses. The interview can be used to assess children and adults as long as their mental ages are two years or higher. Because the ADI–R is an interview rather than a test, and because it focuses on behaviors that are rare in non-affected individuals, it provides categorical results rather than scales or norms.

Inter-rater and test-retest reliability, as well as internal validity, have been demonstrated for the ADI-R.

Reliability

Test-retest reliability and interrater reliability were assessed in unselected clinical (n = 50) and nonclinical (n = 30) populations. Concurrent validity (n = 120), discriminant validity (n = 120), and criterion validity (n = 29) were evaluated in Autistic Spectrum and non-Autistic patients. Results: Test-retest and interrater reliabilities were excellent (most intraclass correlation coefficients > 0.9).

Validity

The factor structure, internal consistency, and convergent validity of the Autism ADI-R algorithm items were examined in a sample of 226 children with Pervasive Developmental Disorders. Exploratory factor analyses indicated a three-factor solution closely resembling the original algorithm and explaining 38% of the variance, with one significant discrepancy. Unlike the algorithm, all nonverbal communication items were associated with the Social factor. Internal consistencies of domain scores ranged from .54 to .84. Correlations between ADI-R domain and total scores and instruments assessing adaptive behavior, psychopathology, and Autism were examined. They indicated some similarities between constructs, but also that the ADI-R measures Autism in a unique fashion.

Concurrent validity (agreement with independent clinician formulation) was very good (mean [kappa] = 0.74). Criterion validity, a comparison with the Autism Diagnostic Interview, was excellent. Discrimination between Autistic Spectrum versus non-Autistic subjects was almost perfect (sensitivity 1.0; specificity > 0.97).