7. Discussion

Chapters 2 through 6 of this report describe the variations across the States in each of the 23 measures for which estimates were produced based on the 2004-2005 National Surveys on Drug Use and Health (NSDUHs), as well as changes between 2003-2004 and 2004-2005 for all measures except for serious psychological distress (SPD) and major depressive episode (MDE). Past reports have noted the regional clustering of States with similar high rates of substance use (e.g., the high rates of alcohol use among northern States in the Northeast and Midwest), the inverse relationship between perceptions of risk of use of a substance and actual use (e.g., between marijuana use and the perceived risk of using marijuana), and the similarity in the ranking of States for related substance use measures (e.g., use of illicit drugs and use of marijuana) (Wright & Sathe, 2004, 2006).

This report marks the first time that comparable State estimates have been available for three consecutive time periods (2002-2003, 2003-2004, and 2004-2005). In this report, statistical tests have been conducted to determine whether there was a significant change between 2003-2004 and 2004-2005. An earlier report tested for differences between 2002-2003 and 2003-2004 (Wright & Sathe, 2006). With both sets of tests, it can be determined whether any State has had a significant change in the same direction for both the earlier and the later period. However, those tests are not able to determine whether the change between 2002-2003 and 2004-2005 is significant or not for a specific measure, State, and age group unless both changes were significant and in the same direction. Because there are State-level random effects in the model that do not vary much from year to year, this can result in a fairly large positive correlation (depending on the measure) between the 2002-2003 estimates and the 2004-2005 estimates. Given the significant size of this correlation, it is important to include it in the tests for significant differences between 2002-2003 and 2004-2005. Thus, a new method for testing is needed and is currently being developed. Based on this method, a web-only supplement to this report will be produced that will compare 2002-2003 and 2004-2005 for all measures that were common to those sets of years. This supplement will include tables with these results and a brief discussion of the results and methodology. Those results are expected to be available in March 2007 on the Substance Abuse and Mental Health Services Administration (SAMHSA) website.

However, some conclusions can be made from the available information in cases where both changes (i.e., between 2002-2003 and 2003-2004, and between 2003-2004 and 2004-2005) have been significant (at the .05 level) and were in the same direction, either increases or decreases. These are discussed in this chapter for a few key measures.

7.1.1 Marijuana Use

Because marijuana is the single most-used illicit drug, it is useful to look at past month use of marijuana. In 2005, of all persons aged 12 or older who had used an illicit drug during the past month, approximately 75 percent had used marijuana (see Office of Applied Studies [OAS], 2006a, Table 1.1). The areas with the highest rates of past month use of marijuana in 2003-2004 among persons 12 or older were States in the Northeast and West (Wright & Sathe, 2006, Figure 2.9). In the Northeast, the States were Maine, Massachusetts, New Hampshire, New York, Vermont, and Rhode Island. In the West, the States in the highest fifth were Alaska, Montana, New Mexico, and Oregon. Also, in 2003-2004 among youths aged 12 to 17, the States ranked in the highest fifth were the same as those among the 12 or older population, with the only difference being the replacement of New York with Colorado (Wright & Sathe, 2006, Figures 2.9 and 2.10).

At the national level between 2002-2003 (6.2 percent) and 2003-2004 (6.1 percent), and between 2003-2004 and 2004-2005 (6.0 percent), no significant changes in marijuana use occurred among the 12 or older population (Wright & Sathe, 2006, Table C.3; Table C.3 of this report). However, there were some decreases among youths and young adults. Among youths, there was a decrease between 2002-2003 and 2003-2004 in the South (from 7.4 to 6.8 percent). Nationally among persons aged 18 to 25, the percentage using marijuana during the past month decreased from 17.2 percent in 2002-2003 to 16.6 percent in 2003-2004. This was accompanied by a significant decrease for this age group in the Midwest. Between 2003-2004 and 2004-2005, the only significant decrease nationally was among youths, from 7.7 to 7.2 percent. This decrease in past month marijuana use was reflected in similar declines among youths in the Midwest and West.

At the State-by-age group level, from 2002-2003 to 2003-2004, there were 20 declines in past month marijuana use across all age groups, and a single increase (Wright & Sathe, 2006, Table C.3). The sole increase was in New Mexico among persons aged 12 or older. That increase resulted from small, but not statistically significant, increases across the three age groups (12 to 17, 18 to 25, and 26 or older), that when combined, caused a significant increase for the combined 12 or older population. Seven of the decreases occurred in the West, six in the South, four in the Northeast, and three in the Midwest. In the 18 to 25 age group, which displayed the only significant decline at the national level during 2002-2003 to 2003-2004 period, two decreases occurred in Western States (Nevada and Washington), two in the Northeast region (New Hampshire and Rhode Island), one in the South (District of Columbia), and one in the Midwest (Ohio).

Between 2003-2004 and 2004-2005, there were a total of 13 State/age group changes during this period, and all of them were decreases (Table C.3). Seven of the decreases were from States in the West: one each in Alaska, California, and Hawaii, and four in New Mexico. One decrease was in Massachusetts in the Northeast region. The remaining changes occurred in the Midwest (Michigan and North Dakota) and the South (Maryland). Of the States contributing to the national decline among youths aged 12 to 17, four of them were from States in the West and two came from States in the Midwest.

Comparing the consistency of changes between 2002-2003 and 2003-2004 with those between 2003-2004 and 2004-2005, none of the changes in past month marijuana use in the first period occurred in the same domains in the second period (Table C.3 of this report; Wright & Sathe, 2006, Table C.3). The decline in the rates of past month marijuana use among persons aged 18 to 25 in the first period (2002-2003 to 2003-2004) was not followed by a similar change in the second period (2003-2004 to 2004-2005). The decrease among youths aged 12 to 17 in the South in the first period was not reflected among youths in the second period. Comparing the 21 State/age group changes from the first period with the 13 changes from the second period, the only change common to both was in New Mexico among persons aged 12 or older. But even here, the direction was opposite. During the first period, New Mexico showed a significant increase from 7.4 to 8.7 percent, while during the second period the change was a significant decrease from 8.7 to 6.6 percent.

It is important to note that this may understate the true number of State-by-age group categories that decreased during both periods because this analysis is defining consistency across both periods only when the changes are both statistically significant (and in the same direction). Other States may display a significant decrease during one of the periods and a nonsignificant decrease in the other period, and tests of the combined decrease from 2002-2003 to 2004-2005 may prove to be statistically significant in those States as well. For example, between 2002-2003 and 2003-2004, New Hampshire had a significant decrease in past month marijuana use among persons aged 26 or older from 7.4 to 6.0 percent (Wright & Sathe, 2006, Table C.3). The comparable New Hampshire estimate for this age group for 2004-2005 is 4.9 percent; however, the decrease between 2003-2004 and 2004-2005 was not significant (Table B.3 of this report). Combined, the total change from 2002-2003 to 2004-2005 is from 7.4 percent to 4.9 percent, and the proposed supplemental testing will determine whether the difference is statistically significant or not.

It is notable that a number of the decreases have occurred in States that were ranked in the highest fifth among persons aged 12 or older. Eight of the States that were ranked in the highest fifth (either in 2002-2003 or 2003-2004) had decreases in one or more age groups either during the first period or the second period (Alaska, Colorado, District of Columbia, Massachusetts, Montana, New Hampshire, New Mexico, and Rhode Island) (Figure 2.9 and Table C.3 in this report; Wright & Sathe, 2005, Figure 2.9; Wright & Sathe, 2006, Table C.3). Across the 20 declines in the first period and the 13 declines in the second period, those States accounted for a total of 16 of the State/age group decreases.

7.1.2 Underage Binge Use of Alcohol

In 2005, the national rate of underage binge alcohol use in the past month was 18.8 percent, which represents about 7.2 million persons aged 12 to 20 (see OAS, 2006a, Tables 2.105A and 2.105B). The States with the highest rates in 2003-2004 in the Midwest were Iowa, Kansas, North Dakota, South Dakota, and Wisconsin (Wright & Sathe, 2006, Figure 3.14). The States with the highest rates in the Northeast were Massachusetts, New Hampshire, and Rhode Island. The remaining States that ranked in the top fifth were Montana and Wyoming. In 2004-2005, the States in the highest fifth were the same with the following exceptions: New Hampshire and Wyoming were replaced by Vermont and Nebraska (Figure 3.14).

There were no significant changes nationally either between 2002-2003 (19.2 percent) and 2003-2004 (19.4 percent) or between 2003-2004 and 2004-2005 (19.2 percent), and no changes among any of the four census regions (Table C.12 in this report; Wright & Sathe, 2006, Table C.12).

There were only five changes among States between 2002-2003 and 2003-2004: Iowa and Oklahoma had increases, while North Carolina, South Carolina, and Tennessee had decreases. Between 2003-2004 and 2004-2005, eight States had changes in the rate of binge use of alcohol among persons aged 12 to 20: Illinois, New Hampshire, New Mexico, North Dakota, South Dakota, and Washington decreased, while Texas and Utah increased (Table C.12 in this report).

Comparing the State changes across the two periods (2002-2003 to 2003-2004 and 2003-2004 to 2004-2005), none of the States that had a change in the first period had a corresponding change in the second period. When the new results on testing for any significant changes between 2002-2003 and 2004-2005 become available, some of the States (e.g., New Mexico, North Carolina, and South Carolina) may indicate a significant decrease for that period.

Of the three States with decreases in the first period and the six with decreases in the second period, only three of them were among those that were ranked in the highest fifth for underage binge drinking, either in 2002-2003 or 2003-2004: New Hampshire, North Dakota, and South Dakota (Wright & Sathe, 2006, Figure 3.14 and Table C.12).

7.1.3 Youth Cigarette Use

In 2002-2003, the percentage of youths aged 12 through 17 who had used cigarettes in the past month was 12.6 percent (Wright & Sathe, 2005, Table B.12). The past month rate of cigarette use among youths declined between 2002-2003 and 2003-2004 from 12.6 to 12.0 percent, and again between 2003-2004 and 2004-2005 from 12.0 to 11.3 percent (Table C.14 in this report; Wright & Sathe, 2006, Table C.14). Among youths, there were similar declines for both periods in the Midwest and South. Both in 2002-2003 and in 2003-2004, the highest rates of past month cigarette use among youths occurred typically in the Midwest and the South (Wright & Sathe, 2005, Figure 4.6; Wright & Sathe, 2006, Figure 4.6). The States that were ranked in the highest fifth among youths in 2002-2003 were Minnesota, Missouri, Nebraska, North Dakota, South Dakota, and Wisconsin in the Midwest; Arkansas, Kentucky, and West Virginia in the South; and Montana in the West.

Over the first period, five States had significant changes among youths, of which four were decreases and one was an increase (in California) (Wright & Sathe, 2006, Table C.14). Of the four decreases, two (Minnesota and Nebraska) were for States ranked in the highest fifth in 2002-2003 (Wright & Sathe, 2005, Figure 4.6). The other two declines were in Florida and Illinois.

During the second period, 2003-2004 to 2004-2005, five States (Alaska, Louisiana, North Dakota, Ohio, and Texas) had decreases among youths (Table C.14). Only North Dakota ranked among the highest fifth in 2003-2004 (Wright & Sathe, 2006, Figure 4.6). Of the States with decreases in the rates of cigarette use in this age group, none of them occurred in the same States that had a decline in the first period. Because there was a decline in cigarette use among youths nationally and in two regions across both periods, the planned future testing to detect significant changes between 2002-2003 and 2004-2005 may reveal that a number of the States mentioned above may have significant declines across the two periods.

7.1.4 Dependence on or Abuse of Alcohol or Illicit Drugs in the Past Year

In 2002-2003, 9.2 percent of all persons aged 12 or older met the criteria for having been dependent on or having abused an illicit drug or alcohol during the past year (Wright & Sathe, 2005, Table B.18). The prevalence rates have remained basically unchanged between 2003-2004 (9.2 percent) and 2004-2005 (9.3 percent) (Table C.20). Because alcohol dependence or abuse rates are higher than rates of dependence on or abuse of illicit drugs, States with high rates of alcohol dependence or abuse tend to report the highest rates of dependence on or abuse of illicit drugs or alcohol. The States that were ranked in the highest fifth for dependence on or abuse of illicit drugs or alcohol in 2003-2004 among persons 12 or older included three States from the Midwest (Wisconsin, North Dakota, and South Dakota), five from the West (Arizona, Colorado, Montana, New Mexico, and Wyoming), one from the South (District of Columbia), and one from the Northeast (Rhode Island) (Wright & Sathe, 2006, Figure 5.17). Eight of the States that ranked in the highest fifth in 2002-2003 were the same as those in 2003-2004, trading Wisconsin and Wyoming for Nebraska and New Hampshire (Wright & Sathe, 2005, Figure 5.17).

Between 2002-2003 and 2003-2004, the rates of past year dependence on or abuse of illicit drugs or alcohol did not change nationally in any age group or region except in the 18 to 25 year old age group in the South (decreased from 20.2 to 19.3 percent) (Wright & Sathe, 2006, Table C.20). There were only five changes among all States and age groups during this period. Iowa and Kansas showed increases among young adults aged 18 to 25 and youths aged 12 to 17, respectively. North Carolina, Ohio, and Rhode Island indicated decreases among young adults aged 18 to 25. Rhode Island was the only State from the top-ranked States in 2002-2003 (Wright & Sathe, 2005, Figure 5.17) among persons 12 or older that had a decrease (albeit, in the 18 to 25 age group) in this period.

Between 2003-2004 and 2004-2005, a national decrease in dependence on or abuse of illicit drugs or alcohol occurred among youths, from 8.9 to 8.4 percent (Table C.17 in this report). During the same period, there were 11 changes at the State/age group level that included 6 increases and 5 decreases. One of the increases was in Montana (in the 18 to 25 age group), a State ranked in the highest fifth in 2002-2003 among persons 12 or older, but 2 of the decreases occurred in New Mexico, another State ranked in the highest fifth (Wright & Sathe, 2005, Figure 5.17). None of the decreases in the first period occurred in the same State/age group as those in the second period.

7.2 Validation

Given the unique NSDUH design and limited availability of independent data sources that provide State-level estimates, it is difficult to validate NSDUH State estimates using external sources. In the past, State estimates from this survey (prior to 2002 named the National Household Survey on Drug Abuse [NHSDA]) have been compared with estimates from the Behavioral Risk Factor Surveillance System (BRFSS) and the Youth Risk Behavior Survey (YRBS) sponsored by the Centers for Disease Control and Prevention (CDC, 2006a, 2006b). However, these CDC surveys (a) did not focus extensively on substance use, (b) employed different data collection methods, (c) did not cover all of the States on an annual basis, and (d) had varying degrees in potential response and nonresponse bias. It is, therefore, difficult to know how much confidence should be placed on comparing the results of surveys that are so different in design and implementation. A recent paper compared past month binge drinking rates among persons aged 18 or older from the BRFSS both at the national and State levels with comparable estimates from NSDUH for the years 1999 and 2001. Miller et al. (2004) found the rates for BRFSS to be lower than those from NSDUH both at the national level and for 46 States although the correlation between the two sets of estimates was fairly high. It was speculated that the differences are due to the perceived greater privacy of the NSDUH that encourages respondents to give honest answers.

Although external validation of NSDUH findings is problematic, internal validation of the State estimates can be useful. Because the State prevalence levels for 2004-2005 are estimated in the same manner as they were for earlier years, the procedures and the results of the validation done for prior estimates apply to these estimates.⁹ The average relative absolute bias (RAB) values from the 2000 State report (produced by pooling the 1999 and 2000 NHSDAs) that compare large sample benchmark values with small sample hierarchical Bayes estimates are as follows (see Tables B.22 to B.25 of the 2000 State report; Wright, 2002b):

past month use of marijuana, 4.07 percent;
past year use of cocaine, 7.88 percent;
past month binge alcohol use, 0.98 percent; and
past month use of cigarettes, 1.22 percent.

These results suggest that, if, for example, the true value of past month use of marijuana for persons aged 12 or older in a State with a sample of about 1,800 persons was 5 percent, the small area estimate would, on average, fall within 0.2 percent (4.07 percent × 5 percent) of the true value. The precision of these estimates is better than that from corresponding design-based estimates of the same sample size. By combining 2 years' data, the prediction intervals are about 25 to 35 percent shorter, depending on the substance (see Appendix B of the 2000 State report, Wright, 2002b).

As noted in past State reports, the models may not adequately adjust for differential nonresponse and bias effects at the State level. Any such bias resulting from nonresponse that varied in relation to the prevalence rates would raise concerns about comparisons among States.¹⁰ For such bias to exist after nonresponse adjustments have been made requires that the true probabilities for persons to respond to the survey still depend to some degree on whether they have used a substance or not.