Validation of the Child Sexual Abuse Knowledge Questionnaire

ABSTRACT

A validation study of the Child Sexual Abuse Knowledge Questionnaire (CSA-KQ) was conducted on a sample of 1712 non-empanelled jurors in the greater Sydney area, Australia. The CSA-KQ contains nine items derived from empirical findings on common misconceptions about typical features of abuse offences, children’s responses to child sexual abuse, and their ability to give reliable evidence. Study 1 tested the factor structure of the questionnaire in a sample of 843 non-empanelled jurors. The best model indicated by exploratory factor analysis had two factors: the Impact of Sexual Abuse on Children and Contextual Influences on the Report. Study 2 cross-validated the findings and tested the predictive validity of the CSA-KQ in a realistic simulated trial in which an 11-year-old complainant alleged abuse by her grandfather. Confirmatory factor analysis replicated the findings of Study 1, showing strong reliability for each of the factors (ρ_y = 0.70 to ρ_y = 0.80) and for the CSA-KQ (ρ_y = 0.76). CSA-KQ scores were significantly correlated with the perceived credibility of the complainant (r = 0.23). Moreover, the CSA-KQ scores predicted verdict: jurors with greater knowledge about CSA were more likely to convict the defendant than jurors who knew less about CSA.

KEYWORDS:

• Child sexual abuse
• juror knowledge
• exploratory factor analysis
• confirmatory factor analysis
• validation study

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Similarly, three-factor model was tested. The model fit statistic was significant, χ²(12) = 22.88, p = .029. RMSEA = 0.033 [0.010; 0.053], CFI = 0.993, TLI = 0.979, SRMR 0.017. The two-factor solution was preferred, as (1) the model fit statistics for two- and three-factor solution did not differ considerably from each other; and (2) Factors 1 and 2 in the three-factor model were statistically (r = 0.65, p < .05) and semantically related, indicating that they measured the same construct.

2. A separate logistic regression was conducted to test the effect of juror gender and education on verdict. Analyses revealed that neither gender (OR = 0.74, 95% CI [0.55; 1.00], p = .048) nor education (OR = 0.87, 95% CI [0.64; 1.18], p = .379) were diagnostically predictive of verdict. Although gender was statistically significant, the p value of .048, if corrected to two decimal places, would no longer be significant, and a relatively large sample size as was used in this study, is more likely to produce a spurious significant effect. As for perceived credibility, controlling for juror gender and education did not change the results of the logistic regression.

Additional information

Funding

This research was supported by funding from the Australian Research Council [grant number DP110103706] to Jane Goodman-Delahunty and Annie Cossins and by University of New South Wales Law School Research Support Funds.