Study 1
Elder and colleagues (2002) used a randomized pretest–posttest control group study design to examine the effectiveness of the Sembrando Salud program compared with an attention-control condition. The study was conducted from January 1996 to December 1999. The Migrant Education Program in San Diego County, California, identified the participants. Twenty-five schools within 17 school districts were eligible to participate; ultimately, participating families came from 22 schools in 15 schools districts. (Two schools did not participate because of academic changes in the school calendar, and the third school was ineligible owing to its small number of migrant families. One district did not wish to share its rosters of migrant family names, and the other had an insufficient number of eligible families.)
Ultimately, 660 adolescents (49 percent females) and one of their adult caregivers agreed to participate in the study. The randomization of the schools resulted in 367 families assigned to the treatment condition and 293 families assigned to the attention-control condition. The participants were predominantly Mexican American. The average family income fell well below the poverty level.
Each condition was designed to be equivalent in all respects (except for the content) and included eight weekly, 2-hour sessions with parents attending three of the eight sessions jointly with their adolescent. Each week was formatted into small-group evening sessions held on school grounds or at a neighborhood community agency. The treatment group received the 8-week Sembrando Salud program. The attention-control condition was a first-aid/home-safety educational program concentrated on preparation for an emergency (such as assembling a first-aid kit). Participants were provided with information on first aid and wound care, bleeding and burn care, and home safety.
Data was collected through face-to-face surveys offered in English and Spanish. Baseline surveys were conducted before program implementation, with parents and adolescents assessed simultaneously in separate areas. The same survey was given to all participants following the completion of the group educational sessions and at 1-year and 2-year postintervention follow-ups. Attrition was similar across the treatment and attention-control condition groups at each follow-up period.
The project-developed survey incorporated previously developed scales and items translated into Spanish and “back translated.” The survey included 201 items and collected information on numerous domains. The primary outcomes of susceptibility to smoking and alcohol as well as 30-day drinking were measured by items developed for the California Tobacco Survey. Adolescents were coded as “susceptible” if 1) they were current smokers or drinkers, 2) they did not show firm resolve not to smoke or drink in the future, 3) they would accept a cigarette or drink from a friend, or 4) they intended to smoke in the next year. In addition, attitudes that were assessed included anticipated outcomes for use, intentions for use, and self-standards (in terms of whether the youth viewed himself or herself as tolerant of the identity of being a smoker). Finally, a six-item scale was used to assess parent–child communication.
Since randomization was based on schools, generalized estimating equations were used to account for the repeated measures and for the effects of clustering that resulted when intervention groups within schools were formed. No subgroup analyses were conducted.