Tests for statistical significance in Table 3 show a lack of parameter differences across substances at age 11. This is not surprising, given the low endorsement rates in this age group, which resulted in large standard errors on location and discrimination estimates. At age 14, substances form two distinct clusters in terms of location. Tobacco and alcohol comprise the less severe group. Greater location parameter estimates and farther right placement of the IRFs for marijuana, other controlled substances, and uncontrolled substances indicate that individuals who used these substances by age 14 would be higher on the poly-substance use trait than those who report only alcohol and/or tobacco use. At age 17, tobacco and alcohol remain the least severe indicators of poly-substance use. The other substances are now more spread out along the underlying dimension, with marijuana being less severe than the other controlled substances item. Use of uncontrolled substances falls roughly between marijuana and other controlled substances in terms of location, but it provides relatively little information (i.e., having a notably lower discrimination value when compared with marijuana and other controlled substances).
