Finally, we probed the associations of parent
education and family structure with smoking
for both groups. Because CW-involved youths
are often of lower socioeconomic position than
are community youths, we wanted to deter-
mine whether CW involvement was serving as
a proxy for disadvantage, or whether distinct
differences existed for CW-involved youths. To
do this, we estimated probabilities of smoking
for each group, matching parental education
at 12 years with single-parent households. For