All variables were eligible to remain in the model based on results from collinearity diagnostics

Unfortunately, the authors were unable to test such associations because no such parenting variables were measured. Also, participants in this study are at risk for an adverse health outcome, as defined by an LTBI diagnosis. To our knowledge, no published literature suggests a link between increased participation in risk behaviors of the variety studied here and an LTBI diagnosis. And, alcohol and tobacco rates reported here are lower than those reported by the Centers for Disease Control and Prevention for Latinos in the same geographical region . Nevertheless, it is unclear what effect this has on the ability for these results to generalize to a larger sample of Latino adolescents. More reliable and valid predictors of alcohol use could be very valuable for clinicians treating adolescents with isoniazid , where an interaction with concurrent alcohol use could severely damage the liver. Based on experiences learned from this study, future studies would benefit from more precise measures of alcohol and tobacco use. The measure used in this study reflects ‘ever’ use, which may be too general. Preferable measures would query specific time periods , quantity or use at each episode and perhaps conditions under which alcohol or tobacco was used . It would also be important to know how often alcohol and tobacco are used simultaneously and if one serves as a prompt for the other. Furthermore,industrial rolling racks future studies along international borders should attempt to control for the social and legal influences related to alcohol and tobacco accessibility and acceptability. The only such measure available in this study proved to be under specified, and therefore was not included in the analyses.

Despite limitations, this study demonstrates clear strengths. First, this sample of Latino adolescents came from an area of high transition as defined by their proximity to a busy international border. This is important, especially considering reports of increased risk behaviors in border communities . Additionally, comparisons between correlates of the two dependent variables reveal a high level of generalizability of the model, as defined on two dimensions. First, the model achieved almost equivalent amounts of explained variance. Second, the peer influence block was theoretically presumed to be most powerful as manifest by its position as the final block. This theoretically proposed relationship was confirmed for both alcohol and tobacco use. Based on findings reported here and in previous research, future studies would benefit from examining the dynamic between parents and peers in preventing alcohol and tobacco use. This should involve discovery of mechanisms for parents to minimize negative peer influences. The best-known risk factors for adolescent use of gateway drugs are related to peer and school influences . An additional dimension sometimes overlooked is that of parenting practices . This dimension may be overshadowed by strong peer and school influences, but some studies report that parenting practices account for as much as 40% of the variance in adolescent’s risk behaviors . A summary of the parenting literature reveals that common errors in parenting practices include inconsistent and poor timing of reinforcement and inappropriate timing and type of punishment . The Behavioral Ecological Model  that guides the current study, indirectly contends that parents should be actively involved in arranging contingencies in their child’s environment.

When used appropriately, these parenting strategies can establish pro-social skills, influence peer relationships and support adolescents to avoid alcohol, tobacco and other drugs . Measurement error is an unavoidable part of research, no matter how reliable and valid the instrument and despite quality control measures . This is especially true when measuring parental influence. In rare instances in the literature when both parents and their children report about parenting behaviors, parents demonstrate a tendency to overestimate their involvement, especially with regards to monitoring and discipline . In fact, multiple studies report correlations between parents’ and children’s reports of parenting behaviors as low as r = 0.11 . Notwithstanding, the convention of asking parents to report about their own behavior is face valid and probably represents one of the most reliable sources, despite the in congruence with the child’s reports . However, because both sources of information are only approximations of the actual events, both should be considered when researching the effects of parenting. One justification for measuring the child’s responses is based on the view that the child’s perception of parenting practices influences his/her behavior . Yet, arguments to the contrary suggest parents are in a better position to report about their own behavior. Assessment of the parents’ practices from both perspectives may be warranted in light of this controversy . A more complete understanding of the parental influence requires an improved understanding of the discrepancies between parents’ and children’s reports of parenting behaviors . Most of the research in discrepant reports between parents and adolescents has not been replicated across cultures and ethnicities. The purpose of this study was to identify whether parent or adolescent reports of parenting practices explains more variance in adolescent gateway drug use, and to compare variables that emerge as significant in the respective models developed for this analysis.

This study was part of a larger intervention study of Latino adolescents with latent tuberculosis infection . For the present study, baseline data from adolescents and one parent were used to create sequential regression models. Two separate models were created, one for parents and a second for adolescents. In each case, adolescents’ gateway drug use was sequentially regressed on 14 independent variables, entered in four blocks: demographics, parental influence, school influence, and peer influence. Blocks were added sequentially to the respective models based on conceptual proximity to the dependent variable, with the most proximal variables added last. The sample of 252 Latino adolescents in this study were high school students ages 13 to 19. They were attending school in south San Diego County, tested positive for LTBI, and volunteered to participate in a medication adherence trial for LTBI treatment. The San Diego State University Institutional Review Board approved the study. Adolescents were ineligible to participate in the study if they had definite plans to relocate from the area within 12 months and/or to receive LTBI treatment in Mexico. After consenting,marijuana drying racks bilingual interviewers interviewed participants and completed baseline self-reported interviews.The main dependent variable of interest was adolescent gateway drug use. Much of the published literature considers gateway drug use to involve alcohol, tobacco and marijuana . A similar approach was employed for this study. Respondents were asked if they had ever used alcohol . This included beer, wine and hard liquor . Respondents were also asked if they had ever used tobacco and finally, if they had ever used marijuana . The three alcohol variables were combined to create an alcohol scale . Next, the alcohol variables were combined with tobacco and marijuana to create a gateway drug use scale . Any use of alcohol resulted in a value of 1, use of alcohol and either tobacco or marijuana was coded as 2, and use of alcohol, tobacco and marijuana resulted in a value of 3. No student reported using tobacco alone. Age, gender, foreign-born status, acculturation, and receiving an allowance were all selected from the demographics section of the adolescent baseline interview. Foreign-born status was ascertained by asking the country of birth. Respondents born outside of the United States were coded as foreign-born. Acculturation was measured using the bidimensional acculturation scale for Hispanics . The acculturation scale consists of 24 questions regarding language use , linguistic proficiency , and electronic media use . Each question had four possible responses, very poorly, poorly, well, or very well. The questions are separated into 2 domains, Hispanic and non-Hispanic , with 12 items in each. For each cultural domain, an average of the 12 items is calculated, obtaining a mean range of scores between 1 and 4. Scores on both domains were used to determine the level of acculturation. Acculturation categories are computed using a 2.5 cutoff score to indicate low or high level of adherence to each cultural domain. An individual scoring higher than 2.5 in both domains was considered bicultural .Demographic variables were entered in block 1 and accounted for a significant 9.0% of the variance in gateway drug use . This block included the parents’ age, parents’ marital status, number of languages the parent speaks and annual household income. Variables reported by the adolescent include age, number of siblings, acculturation, and gender. Age of the adolescent was the only significant variable in this block , with older adolescents being more likely to use gateway drugs.

Block 2, the family influence block was not significant, contributing only 0.8% of explained variance. Variables in this block were only reported by the parent and included celebrating when the child does well, yelling when the child behaves poorly, helping with homework and joining parent groups at the adolescent’s school. Adolescent’s age remained significant with the addition of the second block . The next block represented school influence and was reported by the adolescent. The only new variable added in this block was skipping school in the previous 12 months. This block significantly added 2.5% of explained variance . Skipping school in the past year was significantly related to gateway drug use and adolescent’s age remained significant in the third block . With inclusion of the third block, the model explained 12.2% of the variance in gateway drug use. Block 4 included variables related to peer influence, which included the number of close friends and how many friends used gateway drugs. This block significantly added 21.9% of explained variance . Having more close friends was negatively associated with gateway drug use . Also, peer models of gateway drug use resulted in a significant positive association with reported use . Adolescent’s age was no longer significant in the final model. After entering the fourth block, the full model explained 34.1% of variance in gateway drug use. Table 4 presents the summary statistics for the complete model by block.The purpose of this study was to compare parent and adolescent reports about parents’ parenting behaviors, and to do so in the context of multivariate models for gateway drug use. The BEM guided the selection of variables in each of the two models. Overall, the model based on adolescents’ report explained more variance than the model based on parents’ report. The parenting block, the block defining the key difference between the two models, was only significant in the adolescent’s model. The adolescents’ report of parenting also resulted in more components of subsequent blocks reaching significance, seemingly increasing the precision of he overall predictive model. These discrepancies between parent and adolescent reports suggest a need for additional measures, or more refined measures of parenting practices. The value of the sequential regression approach is tied to its ability to reveal relationships among variables as they are added to the model. As such, brief consideration of each block is warranted. For example, age and skipping school were significantly related to gateway drug use when they entered the model in their respective blocks, but were no longer significant in the final model. After controlling for other variables, the initial association with age and skipping school was essentially mitigated. By contrast, parental involvement at school was not significant initially, but reaches significance as a correlate of gateway drug use in the final model resulting in decreased risk for gateway drug use for more involved parents. When parents report about their own parenting behaviors, none of these variables reach significance. The entire parental influence block is not significant as it enters the model, and in the final model. The final block represents peer influence. Its position as the final block in the sequential regression approach indicates a theory based assumption that it has the strongest association with gateway drug use. The effects of peer influence are well documented in published literature, and results presented here confirm that. Risk for gateway drug use decreases as the social network increases in size as represented here by an increase in the reported number of close friends, and peer use of gateway drugs was strongly related to reported use. The data source for the parental influence block is the defining feature between the two models. As previously mentioned, this block was not significant in the parent’s model. However, when adolescents’ report about their parent’s parenting behaviors the block was significant, both upon entrance in the model and in the final model.