The purpose of this exploratory study was to examine the association between gateway drug use and structural and social environmental variables by gender among a sample of Latino adolescents. The finding that neighborhoods closer to retailers are associated with less favorable neighborhood characteristics has multiple implications. First, it is a confirmation of recent studies, and it extends previous literature in a border county and geographical area with high proportions of Hispanic people. Second, to the extent that retailer data used in this study demonstrated associations consistent with previous findings in the literature it provides evidence of generalizability. The finding that distance to the nearest retailer was positively associated with gateway drug use is puzzling. Gateway drug use among girls was lowest in areas closest to alcohol and tobacco retailers, and remained unchanged for males. This difference observed by gender represents a new finding. One possible explanation relates to parental control and the differential factors that define the gender specific cultures of boys and girls. For example, girls in areas closest to retailers may come under greater parental control to counter anticipated effects of living in a high-risk environment i.e., parents monitor their daughters more than their sons. This may result in lower rates of gateway drug use for females that reside in areas regarded as high-risk. A similar finding was reported by Wahlgren et al. in which high-risk adolescents were less likely than low-risk adolescents to initiate smoking following a counseling intervention, ostensibly resulting from extra attention devoted to prevention efforts as a function of their high-risk status. Continuing with this logic,motel grow racks girls in high-risk neighborhoods, such as those close to retailers, may interact less with their environment.

Young males may not be subjected to the same level of parental control, and may enjoy support to interact freely with their nearby environment. A recent study by Norman et al. demonstrated differential rates of physical activity and interactivity between boys and girls with environmental features such as retailers. Given this context, young males would tend to access nearby retailers via walking in a manner that young females would not. Proximity to nearby retailers may have less influence on females if they are prevented from accessing them. Future studies should be designed to test parents’ differential control of males and females in high-risk environments. This study took place close to the border, one of high transition with one of the worlds’ busiest border crossing . The culture with respect to substance use is more accepting in Mexico compared to San Diego, CA . The increased risk for girls from cohort 2 to use gateway drugs may be an indication of this cross-border influence. Anecdotally, we believe more students from cohort 2 spent significant amounts of time in Mexico, some on a daily basis. Students returning to Mexico alone after school have considerable amounts of unmonitored time while in transit. It is plausible that parents of girls residing in Mexico and attending school in the US were more restrictive of their after school time. The only opportunity in such cases for risk behaviors to occur would be under the influence of the home environment, closer to the border if not in Mexico. This should be the focus of additional research. The design employed in this study enabled a conservative estimate of the relationship between gateway drug use and variables from the structural and social environment. Attempts were made to represent significant theoretical domains, but the analytical model employed here is under specified as many of the variables expected to influence adolescent’s behavior were not available e.g., family members’ use of gateway drugs. Furthermore, measures utilized in this cross-sectional study were not designed a-priori for the questions they are attempting to answer. For this reason, no direct measure of purchasing was included, and the conclusions of this study should not be misinterpreted to suggest that substances were necessarily purchased in areas of shortest distances from residence to the nearest retailer.

Ecological studies in the future should attempt to assess the direct exposure to retailers to increase the likelihood that relationships found at the ecological level reflect individual exposure to retailers. The distinct trends observed by gender provide general support for continued research in this area. The BEM emphasis for future studies includes additional focus on the potential modifying effect of modeling on retailer dispersion. Retailer presence may be nothing more than a marker for substance use modeling, including point-of-sale advertisements i.e., venue for modeling and observing substance use behaviors. Since marijuana is an illegal substance not sold commercially at retailers like alcohol and tobacco, including it as part of the dependent variable begins the process of evaluating effects of this modeling process on substances other than alcohol or tobacco. New models of environmental influences should account for these mediating and moderating processes, as well as other neighborhood features that provide the empirical evidence for modifying retailer policies, public transportation and other means of access to retail sources of alcohol, tobacco or marijuana and test such modifications effect on adolescents’ use of gateway drugs. Several aspects of this study define the significance to the current literature. First, this study employed a novel approach to explore the role of the built environment. Technology utilized in this study will become more precise and easier to use, and early studies based on this technology will provide important foundational work for future research. Second, this study continues important lines of research, with a segment of the population traditionally understudied and integrates an important area of emerging research, the structural environment. Each paper included in this dissertation employs a distinct methodological approach and aims to answer a different question. Notwithstanding their differences, they share a common theme: risk behaviors among Latino adolescents, specifically alcohol, tobacco and marijuana use. The purpose of Paper 1 was to identify correlates of alcohol and tobacco use. Correlate variables were selected based on the BEM and reflect constructs emphasized in learning theory.

The hierarchical model enabled a conservative exploration of the theoretical determinants of alcohol and tobacco use as additional variables were added to the model. After controlling for many other variables from the previous domains, items from the peer domain emerged as significant across both dependent variables, a direct confirmation of the theoretical approach being tested in the sequential regression technique. Paper 2 extended the outcomes to include marijuana use. This involved a creative approach to isolate discrepancies between reports of parents’ parenting behaviors among parent and child dyads, and then to compare the predictive strength of each data source with gateway drug use. Paper 2 fills a gap in the current literature. As reported by Spera , plenty of research has assessed parenting styles, but little has been replicated across cultures, ethnicities, and socioeconomic status groups . Most investigators studying parental influence recognize that adolescent and parental perspectives may be quite different ,rolling grow trays and most studies have relied on only one respondent in the family. The unique features of this study sample provided a venue for comparing the predictive validity between parent and adolescent reports and use of gateway drugs. It was pertinent to the overarching medication adherence community trial to query adolescent respondents and their primary caregiver with respect to parenting practices. Opportunities for comparisons of this variety are rarely available. Results reported here should have a distinct measurement implication. The finding that parental reports were not as predictive as adolescent reports is to be expected. However, interpreting the findings and their implications can be complicated, primarily because the true parenting values remain unknown. Notwithstanding, there are two meaningful implications resulting from these findings. First, the identification of in congruence between parent and adolescent reports and the difference in their predictive validity. These findings provide evidence that these two data sources cannot be considered equivalent. Second, these findings raise questions about the importance of perceptions. For example, adolescent perceptions about their parents’ parenting behaviors may be more important than parental perceptions about their own behaviors. Two possible rationales emerge: 1) parental reporting errors are uniformly overestimates and therefore less believable. It is unclear if parents can maintain objectivity in self-evaluating. They have a high vested interest in being seen as effective parents and their ability to report objectively may be compromised, and 2) adolescent behavioral practices may conform to the perceived desires of the parent according to how much the child perceives intense parental control, restriction, or involvement. Papers 1 and 2 were organized based on adaptations from a landmark publication by Hawkins et al. in which influential adolescent risk domains were identified.

These domains have been corroborated in subsequent research and included demographic related characteristics, family influence, school level influences, and peer influences. Variables used in analyses for papers 1 and 2 were selected based on the BEM, and then organized sequentially into risk domains discussed in the Hawkins paper. Notwithstanding, these risk behavior categories are likely non-inclusive of all risk behavior domains of significance, namely features in the built environment. At the time of their 1992 publication, Hawkins et al. could not have explored the built environment to the same degree of specificity that current methodologies allow. For example, variables from the built environment, like those included in paper 3, would have been much less precise in previous decades. Studies that utilize current technologies and contribute to the body of literature that incorporate such methodologies will be important in years to come. Paper 3 represents an exploratory study, the likes of which may help to shape the way researchers assess previously unstudied risk domains, i.e., the built environment. Despite methodological limitations associated with ecological analyses and the cross-sectional design, results from this study extend our current focus of environmental variables and suggest that we do so in the context of known social predictors of drug use. Findings from papers 1 and 2 were replicated for both males and females.The rational for creating separate regression models for males and females stems from evidence of gender differences regarding interactivity with the built environment. This rational was supported by demonstrating that distance to the nearest retailer was significant for females and not males. Interpreting these findings will almost assuredly be left to future research studies, but ideas are suggested in the last section of this document. The BEM guided this research. While non-prescriptive in nature, the model implicitly calls for the inclusion of variables from multiple domains. Some of these variables were available in complete form, some incomplete variables, and still some variables were unavailable. As a result, the analytical models were in general, probably under specified in the sense that I would have liked to have more, and better, variables that fit the theory. Inasmuch as the data were collected with only some of these analytical purposes in the design, many variables were not present, and still others could be improved upon. All measures in this study were self-report measures collected via interviews, save the variables from the built environment created using GIS software. Self-report measures have inherent sources of error, including recall bias and the opportunity for misrepresenting true values for reasons of social desirability. Notwithstanding their error, improvements could be made to improve their interpretability. For example, dichotomous measures of alcohol use on an ‘ever’ basis do not adequately discriminate use that was purely experimentation versus weekly or daily use of alcohol. While peer modeling emerged as a significant correlate of ever using alcohol , it is also possible that peers influence alcohol use at other frequencies too. That relationship is not clarified in this study. Furthermore, other contextual variables were not included, such as parental use of alcohol, tobacco, or marijuana. Parental use could mediate their adolescent’s to the extent that they model risk behaviors. In general, findings related to parenting should be interpreted in the context of the type and number of parenting items included in these studies. The items included in these papers represent a limited set of parenting items; they may or may not represent all of the parenting behaviors possible, and their generalizability is unclear. For example, joining parent groups at the child’s school may indicate a high level of involvement in other areas also. But the generalizability of this item, and others included in analyses, was not measured.