For evaluating a particular child, T scores are helpful for indicating the degree to which the child deviates from normative samples of peers on each scale. However, for statistical analyses of syndrome scales and DSM-oriented scales, raw scores should be used because the T scores are truncated at 50. For statistical analyses of Internalizing, Externalizing, and Total Problems, either raw scores or T scores can be used because they will usually produce similar results. If different gender/age groups are combined in the same analyses, it may be useful to standardize scores by gender within age groups.
The ASEBA Manuals provide more detailed information about how the norms were derived and the different uses of raw scores and T scores.