Community, Family, and Peer Influences on Alcohol, Marijuana, and Illicit Drug Use Among a Sample of Native American Youth: An Analysis of Predictive Factors

Abstract

Using a cross-sectional sample of 8th, 10th, and 12th grade Native American public school children, this study examines hypotheses pertaining to the ability and influence of measures drawn from social bonding, social learning, and social disorganization theories to account for variations in self-reported lifetime and 30 day use of alcohol, marijuana, and illicit drugs. Results derived from ordinary least squares regression equations show significant associations, most notably with variables from the social learning tradition. In addition, comparisons across equations show significant differences in the impact of the theoretical indicators on substance use between respondents in the sub-samples of those residing on and off reservations. The findings suggest that existing theories offer a promising framework for understanding the process of Native American substance use, and that the role of these in some cases differ for adolescents who reside on and off reservations. Contributions to the literature along with suggestions for future research are discussed.

KEYWORDS:

• adolescent substance use
• Native American
• social bonding theory
• social disorganization theory
• social learning theory

Notes

*Correlation is significant at the .05 level (two-tailed).

**Correlation is significant at the .01 level (two-tailed).

Note: For each variable, the standardized coefficient is shown in the top row and the unstandardized coefficient and standard error (in parentheses) are shown in the bottom row.

*p < .05 (two-tailed).

**p < .01 (two-tailed).

Note: All equations include controls for age, gender, mother/father's education level, and impulsivity.

*p < .05 (two-tailed).

**p < .01 (two-tailed).

Analysis of the 270 cases that were omitted from the Native American sample due to missing or incomplete five-digit zip codes did not show any significant differences (p < .05) when compared to the 1,341 respondents included in the analysis with regard to the demographic variables discussed in Table 1.

All of the variables included in the analysis are measured as standardized scores. Because there are differences in the scale construction of the theoretical variables in the analysis, it was preferable to retain the items as standardized scores. All of the measures are based upon indictors derived from the review of the literature and theoretical overview outlined in the earlier section of the paper. Additional assistance regarding the census level social disorganization indicators was drawn from Ainsworth's work (2002) examining the mediating effects of community context on educational attainment. Confirmatory factor analysis supports each of the items as unidimensional indicators; the reported alpha reliabilities are based on the 1,341 cases that comprise the data for the test of hypothesis one.

Response categories for lifetime reported use of marijuana were given as follows: 1 = 0 occasions, 2 = 1 to 2 occasions, 3 = 3 to 5 occasions, 4 = 6 to 9 occasions, 5 = 10 to 19 occasions, 6 = 20 to 39 occasions, and 7 = 40 or more occasions.

A statistical control for the number of moves the child reported since kindergarten is available in the MPNAS data and was included in the preliminary estimates of the models. Although this indicator was associated with the dependent variables and was included to control for varying lengths of time respondents had lived at their current address, estimates were confounded in the multivariate models due to high levels of colinearity with the residential mobility indicator drawn from the 2000 census. Because the community level indicators are key tenets of social disorganization theory, residential mobility at the zip code level was retained and the number of moves since kindergarten was removed from the multivariate analysis presented in the following section.

Data were screened for missing data, outliers, and other potential data entry errors. Univariate and multivariate examinations of the data supported assumptions of normality, linearity, and homoscedasticity, and colinearity diagnostics, such as variance inflation factor and tolerance scores, showed no evidence of multicollinearity in the final models.

Although not a part of the test of hypothesis two, t tests of the differences in average levels of self-reported use between the on and off reservation respondents were conducted. The results showed that in only one instance (lifetime use of marijuana) were the average scores higher for the on reservation sub-sample than it was for the off reservation sub-sample. In all instances, the differences were very slight and did not come close to approaching statistical significance.