Development of the Computer-Assisted Child Interview (CACI) Tool: Implications for Research and Practice

Background and Purpose: Historically, researchers have been challenged in obtaining reliable and valid self-report data about young children's experiences and perceptions from young children themselves. As such, most have relied primarily on reports from other agents such as parents, teachers, and practitioners even though youth are arguably the “best experts on themselves” (Seita, 2004). Methods for obtaining self-report data from young children have been developed, including Measelle and colleagues' Berkeley Puppet Inventory (1998), but these methods continue to have clear limitations. Meanwhile, the use of computer-assisted technologies in human services has continued to proliferate. The purpose of the Computer-Assisted Child Interview (CACI) project is to develop an assessment tool that is easy and enjoyable for young children to use and, at the same time, is reliable, valid, and expedient for collecting research data. A prototype of the CACI was evaluated for reliability and validity and results are encouraging. However, a number of ethical and practical issues must be considered prior to the implementation of this and other computer-assisted assessment tools including client physical and emotional isolation versus interaction, the potential therapeutic benefits of the assessment process, and the implications of computer and other web-based assessment tools in the context of shrinking mental healthcare resources. This paper describes the results of the CACI evaluation, with special emphasis on these issues.

Methods: Data on CACI reliability and validity were collected from two federally-funded, parallel studies with children aged 4 to 10 years old taking place in the Western (n = 185) and Midwestern United States (n = 391). Internal consistency values for the 11 CACI subscales were calculated for each study by age cohort (i.e., less than 6 years old, 6 years or older). Correlations coefficients among subscales and between subscales and measures of external validity (i.e., parent, teacher, and staff reports of subscale constructs) were also calculated and patterns of concordance and discordance were analyzed.

Results: Except for the Neglect and Hygiene subscale, internal consistency values for other subscales were acceptable for older children, with a range of .64 to .81. For younger children, internal consistency values were also acceptable, ranging from 0.62 to 0.81, with the exceptions of the School Experience (α = 0.55 Western study population; α = 0.54 Midwestern study population) and Peer Relations (α = 0.51 Western study population; α = 0.57 Midwestern study population) subscales. Validity data for all CACI subscales were mixed but, in general, there was a pattern of support for the Externalizing Behavior, Internalizing Behavior, Peer Relations, Parental Discipline, and Parent-Child Attachment subscales.

Conclusions and Implications: The CACI is a promising tool for obtaining reliable and valid self-report data from young children. Among the CACI's many advantages are its portability, ease of use, and appealing design. Prior to broad-based implementation, however, researchers and practitioners need to have a thoughtful discussion about these and other advantages in light of the potential ramifications of the growing use of computer-assisted assessment tools.