|
Article Excerpt
YOUNG PEOPLE TEND TO ENGAGE 1N HIGHER levels of substance use and other risky behaviors from the late teens through the mid-20s than at any other age period. Concerning alcohol in particular, the scientific consensus is that both use rates and alcohol-related problems tend to peak around age 21 (Gfroerer et al., 1997; Johnston et al., 2004; Sher and Gotham, 1999; Wechsler et al., 1995). For instance, according to the Monitoring the Future study, roughly 75% of 18- to 22-year-olds reported having consumed alcohol in the past 12 months (Johnston et al., 2004). Furthermore, roughly one third of young people from this age group reported engaging in heavy episodic drinking (HED) (i.e. having five or more drinks in a row) at least once in the past 2 weeks. According to this study, rates of alcohol use decline after ages 23-24, and rates of HED decline after ages 21-22 (Johnston et al., 2004).
To better understand why young people engage in such high rates of alcohol use and to understand the causes and consequences of such potentially harmful behavior, much research has been devoted to examining individual differences in the development of alcohol-use behavior during emerging adulthood (ages ~18-25; Arnett, 2000). In alcohol-use studies, use has typically been operationalized and studied in terms of several distinct aspects of alcohol-use behavior. Five commonly studied aspects of use are as follows: (1) whether an individual has ever used alcohol within his/her lifetime (ever use) or has used alcohol within the past month or year (recent use), (2) the frequency of alcohol use, (3) the quantity of alcohol consumed, (4) the frequency of getting drunk, and (5) the frequency of HED.
Research to date has uncovered a wealth of information regarding these individual aspects of alcohol use during adolescence and emerging adulthood. The majority of this research has used univariate methods, such as prevalence rate analyses, analysis of variance, cluster analysis, and growth mixture modeling (e.g., Aseltine and Gore, 2000; Bachman et al., 1997; Casswell et al., 2002; Gfroerer et al., 1997; Muthen and Muthen, 2000; Wechsler et al., 1995). Univariate analyses of various aspects of alcohol use are generally performed under the assumption that high levels of a particular aspect of alcohol use indicate high levels of alcohol use overall for each individual. However, several studies suggest that the various aspects of alcohol use are independent to an extent. For instance, Baer et al. (1995) found that the correlation coefficients among typical quantity, typical frequency, and peak consumption (i.e., single greatest amount of alcohol consumed in the past month) among high school seniors ranged from 0.42 to 0.58 and shared, at most, 33% of their variance. Using latent growth mixture modeling, Casswell et al. (2002) found different patterns of change for typical frequency and typical quantity of alcohol use between ages 18 and 26, indicating that frequency of use steadily increased over this age period, whereas quantity of use peaked around age 22. Also using this analytical method, Colder et al. (2002) identified three latent classes of frequency behavior and four latent classes of quantity behavior describing the development of alcohol use between Grades 7 and 12. In their five-class model comprising both frequency and quantity behaviors, frequency behavior was found to increase or decrease much more sharply than quantity behavior over time in two of the classes.
This and other research suggest that frequency and quantity behaviors have a complex association, but the nature of this association is not yet clear. For instance, according to prior research it appears that not all drinkers are "low frequency, low quantity" drinkers or "high frequency, high quantity" drinkers. Some may be "low frequency, high quantity" drinkers or "high frequency, low quantity" drinkers. Furthermore, the association between HED and frequency and quantity behaviors has not been well established. Although heavy episodic drinkers are likely also to be high frequency and high quantity drinkers, it is not certain that high frequency and/or high quantity drinkers are also heavy episodic drinkers.
The fact that there is a complex association among these key aspects of alcohol use indicates that multivariate methods are necessary, both for studying the association among the various aspects of alcohol use and for identifying multidimensional patterns of use. Variable-centered methods such as multivariate analysis of variance and factor analysis can be used to investigate correlations among multiple alcohol-use indicators, whereas person-centered methods such as latent class analysis can be used to study the associations among indicators and to identify common multidimensional patterns of alcohol use exhibited by individuals. More specifically, latent class analysis is typically used to identify distinct multidimensional levels, or classes, of a discrete latent variable based on responses to categorical indicators at one point in time.
Extending beyond the traditional latent class approach, longitudinal latent transition analysis (LTA) can be used to determine the proportion of individuals in each class at several points in time and to examine individuals' transitions among classes over time, reflecting development in the latent variable of interest (e.g., Collins et al., 2000; Lanza et al., 2003). For example, Jackson et al. (2001) used information regarding the frequency of alcohol use in the past month, frequency of getting high from alcohol in the past month, and frequency of getting drunk in the past month to build a developmental latent transition model of alcohol use. Using longitudinal data spanning three measurement waves between ages 18 and 24 years, Jackson et al. found that young adults progressed through four multidimensional latent classes of intensifying alcohol use over time: abstainer (low probability of drinking alcohol, getting high on alcohol, and getting drunk), limited-effect drinker (high probability of drinking alcohol but low probability of getting high or drunk), moderate-effect drinker (high probability of drinking alcohol and getting high but low probability of getting drunk), and large-effect drinker (high probability of drinking alcohol, getting high, and getting drunk). Jackson et al. found that alcohol use was more likely to increase between ages 18 and 21 than after age 21 and was more stable between ages 21 and 24 than before age 21. In another study that used ETA to examine the development of alcohol use longitudinally across elementary school (ages 10-12), middle school (ages 13-14), and high school (ages 15-18), Guo et al. (2000) identified four types of alcohol use based on information regarding ever use of alcohol, current use (i.e., alcohol use in the past month), and the occurrence of HED in the past month: nonuse, initiation only, current use only, and HED. Guo and colleagues found that increases in participants' alcohol use were more likely to occur between middle school and high school than between elementary school and middle school.
The purpose of the present study is to complement the findings of Jackson et al. (2001) and Guo et al. (2000) by offering a somewhat different conceptualization of alcohol use based on four aspects of...
|
