Within each age-by-sex group, the IIFs in Figure 1 are summed together to form the Test Information Curves (TICs), shown in Figure 2. From the TICs, we are able to calculate the Standard Error of Measurement (SEM) for any given trait level, also shown in Figure 2. At age 11, the measure of diversity of substances used provides very little information, as evidenced by the relatively low height of the TICs at this age at low to mid trait levels, and the high SEM at trait levels less than two standard deviations above the mean. More information is provided at ages 14 and 17, although at these ages the measure appears to be estimating latent trait levels from the mean to two standard deviations above the mean (at age 14) or within one standard deviation of the mean (at age 17) relatively well. Less information is provided at more extreme trait levels, where the SEM is more substantial. As a result, we would expect trait level estimates for individuals outside of the mid-level trait range to be relatively less reliable.
