Child sexual abuse allegations: Reliability of criteria-based content analysis.

Statement Validity Analysis (SVA), an assessment system for the credibility of children's allegations of sexual abuse, has recently undergone quantification and empirical testing. Twenty-three videotapes of investigative interviews of confirmed child sexual abuse victims were obtained for an interrater reliability analysis. Each tape was rated by two of four trained raters according to the criteria-based content analysis (CBCA) procedure of the SVA method. The average proportion agreement (.75) was equivalent to the value(.72) reported by Steller (1989), but the chance-corrected reliabilities of the CBCA criteria varied from moderate to low (Maxwell's RE coefficient of agreement range =−.22 to 1.00). The criteria need more explicit, behaviorally anchored definitions, and rater training needs to focus on the identified problem criteria. Potential limitations of interpretation include the small restricted sample, the use of uncontrolled community interviews, and the rating of videotapes rather than transcripts of the interviews.