FIGURE 2-1 Ranges in data on the prevalence of mental, emotional, and behavioral disorders among young people.

NOTE: Lines represent the range of estimates from different studies. Boxes represent the interquartile range.

SOURCE: Based on a meta-analysis for the committee by Alaattin Erkanli, Department of Biostatistics, Duke University. A list of the data sets used in the meta-analysis is in Appendix B, which is available online.

of estimates is generally much wider than the lowest 25 percent range, indicating that a few studies tend to generate much higher estimates than do the majority. Several factors contribute to the variability in prevalence estimates: (1) changes in the taxonomy or definitions and criteria used for disorders in different versions of the DSM and the ICD, (2) the evolution of assessment tools over the past few decades, and (3) differences in the populations sampled and the inclusion and exclusion criteria used. For example, since different disorders have different onset ages (see the section on incidence below), samples with different age ranges will show different prevalence rates for many disorders. A fourth factor is that, in surveys of young people (but rarely in surveys of adults), it is normal to collect infor-