Positive Action (PA) has been evaluated in studies in Florida, Hawaii, Nevada, and other States to measure changes in student behavior and achievement at schools implementing the program.
Beets and colleagues (2009) assessed the effectiveness of a 5-year trial of the PA program using a matched-pair design with 20 Hawaii public schools randomized to intervention (n= 10) and control (n= 10). Schools were identified in 2000 and were eligible if the school a) had at least 25 percent of its students receiving free or reduced-price lunch, b) was in the lower three quartiles of SAT scores in the State, c) was located on Maui, Molokai, or Oahu, d) was a K–5 or K–6 public school, and e) had annual student mobility rates below 20 percent.
The 111 schools that met eligibility criteria were classified into 19 strata, on the basis of an index that included such factors as school size, student ethnic distribution, and indicators of student behavioral and school performance. At the 20 participating schools at year 5, the self-reported ethnicities of fifth grade students were primarily:
- Hawaiian or part Hawaiian (26.1 percent)
- Multiple ethnic backgrounds (22.6 percent)
- Non-Hispanic white (8.6 percent)
- African American (1.6 percent)
- American Indian (1.7 percent)
- Other Pacific Islander (4.7 percent)
- Japanese (4.6 percent)
- Other Asian (20.6 percent)
- Other ethnicities (7.8 percent)
- Unknown (1.6 percent)
Control schools continued without making any substantial social or character development program reforms. Intervention schools received approximately 35 hours per school year of sequenced PA lessons. Self-reports of substance use, violence, and voluntary sexual activity were collected on 1,714 fifth graders. Teachers of participating students reported on student substance use and violence. Data was analyzed using t–tests and logistic regressions as well as generalized linear latent and mixed-model methods. The authors indicated limitations, including those resulting from the low prevalence of negative behaviors and the collection of data from fifth graders only for this schoolwide prevention program.
Li and colleagues (2011) assessed the impact of PA in a group of Chicago, Ill., public schools on problem behaviors such as substance use, serious violence-related behavior, current bullying, and disruptive behaviors. A matched-pair randomized control design was used so that schools with similar characteristics were matched, and then one school was randomized to the treatment group. The pool of possible Chicago public schools totaled 438, of which 68 met inclusion criteria. Representatives from 18 of these schools agreed to participate. Fourteen elementary schools eventually participated in the study—seven in the control group, seven in the treatment group. Forty-nine percent of the Chicago public schools population was African American, 39 percent were Hispanic, and 75 percent of the Chicago public schools students were eligible for free or reduced-price lunch.
The program was implemented in 2004–05 with third graders, and they were followed through the 2006–07 academic year. Five assessments were made over the course of the study: a baseline assessment in fall 2004, then follow-up assessments in spring 2005, fall 2005, spring 2006, and spring 2007. Measures of negative behaviors, psychological status, and positive behaviors were included.
Approximately 590 third graders finished the baseline assessment, and 510 fifth graders completed the final assessment (a little more than half—57 percent—were part of the original sample). The students reported their ethnicity at 46 percent African American, 27 percent Hispanic, 7 percent white non-Hispanic, 3 percent Asian, and 17 percent other or mixed.
Measures included lifetime prevalence of substance use and serious violence-related behavior (which were collected through questions on a researcher-developed survey) and bullying and disruptive behaviors (which were collected using the Aggression Scale and the Frequency of Delinquent Behavior Scale). Covariates included age, gender, ethnicity, and measures of baseline problem behaviors. Three-level overdispersed Poisson models were used for analysis to account for nesting (students within schools within school pairs).