Reliability
Questions from the original version of the ADI that were found, through research, to be unreliable or not applicable were removed when the interview was revised. The ADI-R has also been tested thoroughly for reliability and validity using inter-rater reliability, test-retest reliability and internal validity tests. The results of this research have led to the ADI's acceptance among both researchers and clinicians for decades. The ADI-R is often used in conjunction with other related instruments to determine an autism diagnosis.
The writers have published psychometric results that indicate both reliability and validity of the ADI-R. Both inter-rater reliability and internal consistency were good across all behavioral areas investigated in the interview. The interview was also found to have adequate reliability across time. Research comparing ADI-R results of autistic children and children with other developmental disorders suggested that individual questions on the interview were slightly more valid when discriminating autism from mental retardation than the algorithm as a whole. However, further research has led to overall acceptance of the ADI-R algorithm.
Read more about this topic: Autism Diagnostic Interview-Revised