Method:

The study was carried on with an ex post facto, correlational design, and with a non-probabilistic sample of 1,019 subjects. Sample. The sample size was estimated considering the possibility of selecting a subsample of drug users and comparison subjects for a post-stratified analysis, assuring a statistical power of 80% and adequate sensibility and stability. Therefore, this work includes the performed analyses with a sample of 75 drug users and 75 non users, paired by gender, age, school grade and occupation. Instrument. The instrument was a self-applied questionnaire specially developed for the study, according to information obtained in previous focal groups interviews with high school students. The questionnaire showed a global realiability of 0.9154 and between 0.62 and 0.94 in each one of its scales, which included: behavioral beliefs (0.9121), attributed value to behavioral beliefs (0.7964), normative beliefs (0.6480), subject’s disposition to adjust to normative expectations (0.8564), descriptive norm (0.6254), drug use opportunities (0.8129) and perceived behavioral control coping with such opportunity situations (0.9442). A factorial analysis of principal components yielded 16 factors of at least three items each, with factorial weights higher than 0.4, and closely attached to Theory of Planned Behavior’s variables, with an explained variance of 59%.

Analysis. Previous to data analysis, normality tests (Kolmogorov- Smirnov) were performed, indicating the necessity to apply nonparametric tests of differences and to transform the data to be adapted to the requirements of later parametric analyses.

A correlation analysis was carried out to prove the association between behavioral intention and drug use, as well as between the different components of the Theory of Planned Behavior. Finally, linear and logistic regression analyses were conducted to determine the explicative potential of the model and the predictive weight of each variable on the model with regard to the behavioral intention and the consumption of drugs.

Findings:

According to the Mann-Whitney test, compared with students who had not used drugs, subjects that used them at least once in their life showed more favorable attitudes toward consumption (median= 6.9 vs. 3.9, z=-5.22, p=0.000), perceived more social tolerance (median=3.8 vs. 3.5, z=-2.27, p=0.023), were more willing to give in to social pressure for using substances (median=2.0 vs. 1.0, z=-5.598, p=0.000), perceived a higher number of users among their significant others, and less negative consequences Salud Mental, Vol. 30, No. 1, enero-febrero 2007 69 of drug use in themselves (median=16.3 vs. 7.1, z=-4.246, p=0.000), and felt less capable of behavioral control when coping with opportunities for consumption, which, in turn, are more frequent in their case (median=5.7 vs 1.8, z=-6.76, p=0.000).

The correlation between the intention and the behavior of drug use (r=0.41, p<0.000) was allocated inside the range reported in other populations. Drug use intention correlated with attitude toward drug use at r=0.45 (p=0.000), with subjective norm, including additional components at r=0.48 (p=0.000), and with perceived behavioral control at r=0.59 (p=0.000). Drug use correlated with attitude at r=0.51 (p=0.000), with subjective norm at r=0.28 (p=0.001), and with perceived behavioral control at r=0.37 (p=0.000).

Linear regression analysis yielded that the model explained 34% of the variance of drug use intention, which increased to 38% when adding personal and descriptive norm elements to the subjective norm construct. Behavioral control (measured on the basis of the product of exposition to drug use facilitating situations punctuations by perceived behavioral control to cope with these situations punctuations) was identified as the best predictor of drug use intention (B=0.32, p=0.001), followed by attitude toward drug use (B=0.24, p=0.004) and subjective norm, which originally showed a non-significant effect but increased its predictive weight when additional elements were added (B=0.24, p=0.004).

According to the logistic regression analysis, behavioral control is also the best predictor of illicit drug use on the model (odds ratio= 1.42, p<0.000). On the contrary, subjective norm (including personal and descriptive norm) and attitude (odds ratio=1.144, p=0.06) were not significant predictors of drug use.

Discussion:

In general, this findings indicate that the Theory of Planned Behavior showed an acceptable predictive capacity (similar to that found in other populations), and can be taken as a valid theoretical ground to develop a preventive intervention directed to Mexican students of high school education.

As is the case with other populations, the variable in the Planned Behavior Theory with more predictive weight was perceived behavioral control, followed by attitude to drug use and, in third place, by subjective norm. Despite it could be supposed that subjective norm would have a higher predictive weight in Mexican teenagers, findings probably reflected idiocentric and individualistic tendencies reported in other studies.

Results also point out to the convenience of including it in the program of intervention informative components to produce an awareness effect and an impact in the intention of using drugs. Still, they indicate above all, the need to integrate components directed to the development and reinforcement of behavioral control abilities that have an effect in the use of drugs itself.

Specifically, it is considered the convenience of including components for the development of group pressure resistance abilities and assertive communication, appropriate to the contexts in which young people face drug abuse risk situations.