Reciprocal Effects of Substance Use and Violence Perpetration Among Mexican Adolescents: A Gendered Analysis

Background and Purpose:  Prior research suggests substance use and violence perpetration often co-occur among adolescents.  Theoretically, substance use leads to violence perpetration through a disinhibiting effect.  Substance use is also posited to be used as a coping strategy to manage distress associated with violence perpetration.  However, the majority of studies have considered only the unidirectional effect of substance use on violence perpetration.  Few studies have examined the reciprocal effects of substance use and violence perpetration longitudinally.  This study helps fill that gap by examining the bidirectional effects of alcohol use and violence perpetration among a sample of Mexican early adolescents.  Because substance use and violence perpetration tend to be gender-specific, these relationships are examined separately by gender. 

Methods:  This study used three waves of data (N=4,830) from a feasibility trial of a school-based prevention curriculum in Mexico.  Data were collected from middle school students (Time 1 M_age=12.0, SD=.58, Female=48.6%) at 13 schools in Mexico City, Guadalajara, and Monterrey.  The bidirectional effects of 30-day alcohol frequency and two forms of violence perpetration— bullying/aggression and violent criminal behavior— were examined using two cross-lagged path analytic models in Mplus.  Models were estimated separately by gender and controlled for age, site, and treatment condition.

Results:  Global fit indices indicated the model with bullying/aggression had good fit to the data (χ²(4) = 1.91, p = .75; CFI=1.0; RMSEA=.000; SRMR=.004).   Only the model for girls had significant bidirectional paths.  Time 1 and 2 alcohol frequency predicted bullying/aggression at Times 2 and 3, respectively (β = .08, p < .01; β = .07, p < .05).  Time 1 bullying/aggression predicted Time 2 alcohol frequency for girls only (β = .09, p < .001), but Time 2 bullying/aggression predicted Time 3 alcohol frequency for both genders (β_girls = .07, p < .05; β_boys = .06, p < .05).  

For the violent criminal behavior model, global fit indices indicated the model had good fit to the data (χ²(4) = 6.29, p = .18; CFI=.99; RMSEA=.015; SRMR=.009).  Only the model for boys had significant bidirectional paths.  Time 1 and 2 alcohol frequency predicted violent criminal behavior at Times 2 and 3, respectively (β = .08, p < .05; β = .10, p < .05).  Only violent criminal behavior at Time 1 was related to alcohol frequency at Time 2 (β = .12, p < .01).  Additional multi-group analyses will be conducted to formally compare gender-specific models.

Conclusions and Implications:  Findings suggest substance use and violence perpetration mutually influence one another.  However, the type of violence and gender of the individual are important factors to consider.  For girls, bullying and aggression led to and resulted from alcohol use, whereas for boys, violent acts that were more criminal in nature were relevant to understanding the origins and risks of alcohol use.  Findings have important implications for adolescent substance use and violence prevention.