The measurement of children's exposure to violence: A Rasch analysis

Current methods for assessing children's level of exposure to violence (ETV) generate non-linear scores whereas linear, interval level measurement is required in order to make valid inferences about group differences or change across time. Ordinal items can, however, be converted to a linear scale through the use of item response modelling. ETV response data from a demographically diverse sample of forty-nine subjects (29 F, 20 M) ranging in age from 9 to 24 years (M = 17.6 years) was examined for its fit to the Rasch item response model for rating scales. The Rasch model constructs a linear measure for ETV by using ordinal response data as realizations of the probabilities of item endorsement given the level of item extremity and the exposure level of the person taking the item. The resulting person measures and item calibrations are on the same scale. Results show that these data fit the Rasch model quite well. In addition, the hierarchical ordering of item difficulty is consistent with our expectations and, thus, provides some evidence for construct validity of the scale. The experience with ETV suggests that item response models such as Rasch may profitably be applied to other multi-item assessments of social behaviour and experience.