3
Measures
Alcohol and marijuana problems
At each wave, respondents completed a self-report inventory of negative
consequences experienced while or as a result of drinking or using marijuana. Items
that corresponded to specific diagnostic criteria were used to construct a proxy
measure for each symptom and reflect the number of negative symptoms experienced.
Sensation seeking needs
This includes measures of experience seeking and disinhibition, from
Zuckerman’s Sensation seeking form (1979) and novelty seeking from Cloninger’s
Tridimensional Personality Questionnaire (1987).
Family history of alcoholism
Respondents’ and parents’ reports of alcohol abuse in first and second degree
relatives at T1-T3, and standardized information obtained at T4 about family history of
substance abuse from the Family History Research Diagnostic Criteria (Andreason et
al., 1977, 1986) were used to create an indicator of family history positive (FH+) for
alcohol abuse among first and second degree relatives.
Methodology
A growth mixture model approach was used to develop trajectories of problem
drinking. This mixture model method is a semi-parametric group-based technique that
allows for cross-group differences in the shape of developmental trajectories (see Hill et
al., 2000; Muthén and Shedden, 1999; Roeder et al., 1999; Nagin and Tremblay, 1999).
This modeling approach is available through a customized SAS macro (known as Proc
Traj) developed by Jones and colleagues (2001).
To take advantage of the cohort sequential design in modeling the trajectories,
seven ages were scored for problems of alcohol and marijuana use (i.e., ages 12, 15,
18, 21, 24/25, 28 and 30/31). The number of participants included for each use
measurement varied depending on age cohort and time of measurement. For example,
number of symptoms at age 12 was assessed in the youngest cohort (at T1), whereas
two cohorts supplied information about symptoms at age 15 (the middle cohort at T1
and the youngest cohort at T2). The oldest cohort was 24 at T3 and the youngest
cohort was 25 at T4; their scores were combined to provide a measure of alcohol-
related symptoms at age 24/25. The youngest cohort was 30/31 at T5 and their scores
were combined with the oldest cohort, age 31 at T4.
Results
Convention