| ![]() | |||||||||||
![]() |
|
|||||||||||
Youth Substance Use: State Estimates From the 1999 National Household Survey on Drug Abuse |
7. DISCUSSION
This report has examined the use of alcohol, tobacco, and marijuana among youths aged 12 to 17 in each of the 50 States and the District of Columbia in 1999, including risk and protective factors associated with use and the sequence of initiation of alcohol, cigarettes, and marijuana. State estimates that were examined include both model-based estimates and sample-based estimates. The latter were restricted to scales and other continuous data that result in more precise estimates than questions having just "yes" or "no" answers. Still, the smaller sample sizes in the 42 States and the District of Columbia and the wide range of responses for some of the variables resulted in estimates with sizable sampling errors.
The estimates in this report are useful in obtaining a more complete picture of an individual State's substance use problem. In general, the States differed in substance use and related measures. Differences in these measures between States may provide some insights into the reasons that some States have high levels of substance use and others have low levels.
At the State level, there was a high correlation between the small area estimation (SAE) model-based measures for specific substances. For example, there was a strong correlation between State rankings of past month alcohol use and past month binge alcohol use. (Nine States were high for both.) There was also a strong negative relationship between past month alcohol and perceived risk of binge use (six to seven States were the same). There was little relationship at the State level between average age at first use (AFU) of alcohol and past month alcohol prevalence.
The rankings of States on past month use of tobacco were highly correlated with those for past month use of cigarettes. Eight of the States in the top quintile were the same for both substances. The perceptions of low risk of cigarette use was strongly associated with high rates of both cigarettes and tobacco use. There was not a strong relationship at the State level between age at first use of cigarettes and past month cigarette use.
Rankings for past month use of marijuana were moderately correlated with rankings for incidence of marijuana. Five of ten States were in the high group for both. There was a fairly strong relationship between incidence of marijuana use and perceived risk of marijuana. There was also a fairly strong relationship between age at first use of marijuana and past month use of marijuana.
In analyzing the relative standing of a State with respect to these variables, it is important to consider the size of the prediction (or confidence) intervals and the relative spread of the distribution. Some of the variables, such as the average age at first use of marijuana, were based on relatively small samples and had large variations in responses, resulting in rather large confidence intervals. Other variables, such as antisocial behavior, exhibited a narrow range of State estimates, making it hard to make statements about the relative position of a State. Because the true average State level of a specific variable can be somewhat obscured by the sampling error, one should be cautious in dismissing the possibility of a strong relationship even when the correlation among related variables appears to be weak. One should also consider the fact that some of the variables are composite scales based on a number of substances, such as alcohol, cigarettes, and marijuana, and are not substance specific, making inferences about a single substance difficult.
For these reasons, although it is reasonable to consider the possibility of a relationship between any two variables, one should attempt to confirm this (a) by looking at estimates for other age groups for the same substances (included in this report); (b) by looking at estimates for other substances that may be related (these are currently available on the SAMHSA website); and (c) by supplementing the data with other State sources of information. If, for example, the estimates for the age groups 18 to 25 and 26 or older are similarly high, this can confer some assurance that the estimate for ages 12 to 17 is reasonable.
State estimates for risk and protective factor construct scales did not show much variation. Therefore, it is unclear at this point whether the variations in these scales are capable of explaining the differences among States. One of the issues is that some of the factors represent alcohol, tobacco, and other drug (ATOD) scales. However, we know from earlier results that some States (e.g., Kentucky) are high in one substance (such as cigarettes), but low in other substances, such as marijuana. The forthcoming report on risk and protective factors based on the 1999 NHSDA may shed more light on this.
It is beyond the scope of this report to try to interpret the set of data for any State. The focus has primarily been restricted to comparisons of two variables at a time. However, it is important for each State to analyze what the collection of estimates may say about its substance use problem. In Exhibit 7.1, some of the information in the report for the top one fifth of the States has been summarized for past month use of marijuana. Estimates for other variables related to substance use are also presented for these States. The exhibit also includes the national average and the range of State estimates for each variable in order to put the State estimates into perspective. Each variable is labeled with an H for HIGH whenever the State was in the highest category. It is often the case that States may not have had an H for a variable, but were quite close to that level, so this needs to be taken into account. For age at first use, an H was assigned when a State had an average that was in the lowest quintile. If States were tied for the high ranking, those States were conservatively excluded from the high ranking.
Other Hierarchical Bayes estimates have been generally excluded from Exhibit 7.1 because there is a degree of "built-in" correlation among those variables in that the models draw from the same set of independent variables and the same variables are often in a number of models. Two exceptions were the inclusion of two variablesincidence of marijuana and the perceived risk of using marijuana once a month. The sample-based estimates for a number of variables were included even though at times the sampling error may have been large, the range of estimates small, or the scale not specific to the substance.
Exhibit 7.1 Estimates of Past Month Marijuana Use in Comparison with Other Related Items Among the States with the Highest Past Month Marijuana Rates: 1999
Past Month Use1 |
Average Annual Incidence1 |
Perceived Risk Marijuana1 |
Average AFU Alcohol |
AverageAFU Cigarettes |
Average AFU Marijuana |
Difficulty Obtaining Marijuana |
Antisocial Behavior |
Favorable Attitudes to Drugs |
Friends' Attitudes on Drugs |
Peer Drug Use | |
National |
7.4 |
6.3 |
37.3 |
15.7 |
15.0 |
16.2 |
3.41 |
1.15 |
1.55 |
1.61 |
1.69 |
DE |
13.9 H |
7.9 H |
32.3 |
15.6 |
14.4 H |
16.2 |
3.70 H |
1.21 H |
1.73 H |
1.73 H |
1.84H |
MA |
11.9 H |
8.7 H |
29.3 H |
15.6 |
14.6 |
16.2 |
3.53 |
1.17 |
1.57 |
1.60 |
1.77 |
NV |
11.6 H |
8.0 H |
29.8 H |
15.1 H |
14.1 H |
15.1 H |
3.67 H |
1.15 |
1.65 H |
1.64 |
1.79 |
MT |
11.4 H |
7.1 |
27.8 H |
14.8 H |
14.4 H |
15.1 H |
3.47 |
1.19 H |
1.62 H |
1.66 |
1.79 H |
RI |
10.8 H |
7.0 |
33.9 |
14.9 H |
14.2 H |
16.3 |
3.42 |
1.18 H |
1.58 |
1.63 |
1.69 |
NH |
10.7 H |
7.2 |
29.9 H |
15.5 |
16.6 |
16.5 |
3.48 |
1.12 |
1.46 |
1.55 |
1.67 |
AK |
10.4 H |
7.8 H |
32.8 |
15.3 |
14.4 H |
16.0 |
3.49 |
1.18 |
1.53 |
1.59 |
1.72 |
CO |
10.3 H |
7.5 H |
27.0 H |
15.1 H |
14.2 H |
15.9 |
3.70 H |
1.15 |
1.70 H |
1.78 H |
1.80 H |
MN |
9.9 H |
6.6 |
32.8 |
15.6 |
14.8 |
15.6 H |
3.55 |
1.19 H |
1.61 |
1.66 |
1.74 |
WA |
9.9 H |
7.3 |
31.4 H |
15.4 |
15.1 |
15.8 H |
3.52 |
1.16 |
1.53 |
1.61 |
1.70 |
Range |
5.2-13.9 |
4.5-8.9 |
25.8-45.7 |
14.8-16.5 |
14.1-16.6 |
15.1-17.1 |
3.09-3.70 |
1.10-1.21 |
1.30-1.73 |
1.34-1.78 |
1.46-1.88 |
NOTE: The letter "H" placed after an estimate indicates when a State has an average in the top quintile for that item, except for the three age at first use (AFU) items, where the estimates are designated with an "H" when a State has an average that is in the lowest quintile for that item. 1 Estimates presented in these columns are model-based estimates. Source: SAMHSA, Office of Applied Studies, National Household Survey on Drug Abuse, 1999 CAI. |
Some States were in the highest group for a number of variables other than past month marijuana. Delaware was high in 7 of the 10 variables; Nevada for 7; Montana for 7; and Colorado for 8. For the remaining States, Massachusetts had 2 high; Rhode Island had 3; New Hampshire had only 1; Minnesota had 2; Alaska had 2; and Washington had 2.
For all of the 11 measures, the majority of the 10 States had (point) estimates that were "worse than" the national average, although statistically some of the State estimates would be indistinguishable from it. The measures were generally consistent with each other. New Hampshire was an exception to the above, having estimates that were "better" than the national average for 5 of the 11 variables. This would seem to indicate that those variables would not be responsible for its past month prevalence of marijuana ranking in the top quintile.
In analyzing the 1999 data, researchers may find it useful to look at other possible scale measures. For example, a "substance dependence scale" was created by assigning a "zero score" to each youth who indicated that he or she had not used the substance in the past year. However, those records could also have been eliminated from the measure, resulting in an average dependence score of only those youths who had used the substance in the past year. Thus, the measure would just be applicable to past year users in that State and not comparable to other States.
Because the sample sizes for the small States were relatively small for some of the variables, researchers may find it helpful to combine 2 or more years of data in order to obtain better precision for the sample-based estimates. Using both the 1999 and 2000 data, sample sizes of youths for those States would be approximately 600.
The data indicate that the sequence of initiation of cigarettes, alcohol, and marijuana can vary widely among persons, among different subgroups at the national level, and across States. Based on a single years' data, it is difficult to accurately assess how different the patterns are among States, and it may be necessary to combine 2 or more years of data to obtain more precise estimates. The fact that there are a number of variations in the pattern of initiation among the large States indicates a distinct possibility of similar differences among the small States. Because of the stability of the national estimates, knowledge that a State is significantly above or below the national average should be instructive.
The variety of different patterns of substance initiation, ages at first use, and lags between first use of alcohol or cigarettes and later marijuana use at the national level and other levels do not seem to support the hypothesis that either alcohol or cigarettes acts as a "biological trigger" for marijuana use at a fixed age or after a fixed period of time. However, other research has shown that youths who have used cigarettes or alcohol are more likely to have used marijuana compared with youths who have not used cigarettes or alcohol (Kandel & Yamaguchi, 1993; Kandel et al., 1992). The fact that the relationship between cigarette initiation and use and later use of marijuana can vary widely seems to imply that factors other than cigarette or alcohol use may play a large role in determining if and when that use is associated with marijuana use.
![]() This page was last updated on June 03, 2008. |
* Adobe™ PDF and MS Office™ formatted files require software viewer programs to properly read them.
Click here to download these FREE programs now
Highlights | Topics | Data | Drugs | Pubs | Short Reports | Treatment | Help | OAS |