A Randomized Trial of the Self-Management Training and Regulation Strategy (STARS): A Selective Support Program for Students With Disruptive Behaviors
Method: The impact of STARS was assessed using a cluster-randomized, pre-post effectiveness trial. The sample was derived from a nonprobability, convenience sample of fourth and fifth grade students in six elementary schools. Using a universal approach to screening 768 students, 20% of students with elevated levels of disruptive classroom behavior were identified. Following consent procedures, students (i=108) were randomized at the classroom level (j=42) to either the STARS (i=60; j=23) or control (i=48; j=19) conditions. Using assessments of disruptive behavior (α=.95), authority acceptance (α=.93), cognitive concentration (α=.84), perceived autonomy (α=.95), student-teacher relations (α=.72), and social competence (α=.89), data were collected from teachers and students. Following pretest assessment, school-based practitioners provided STARS students with 2-weeks of training followed by 6 weeks of behavioral self-monitoring. STARS students meet weekly to examine self and teacher monitoring data. Control students received routine services. Posttest assessments were collected following allocation. Associations were examined at the α=.05 level using multilevel models to control for clustering. Direct, moderated, and mediated program effects were estimated at the classroom level (Raudenbush & Bryk, 2002). Fidelity, feasibility, and social validity were monitored.
Results: Results indicated students randomized to the STARS and control conditions were equivalent on all observed demographics and outcomes at baseline. Multilevel models suggested STARS was associated with posttest reductions in teacher-rated disruptive classroom behavior (ES=.46) and improvements in social competence (ES=.55), authority acceptance (ES=.47), and teacher-rated (ES=.39) and trending student-rated (p<.10) student-teacher relations. No significant moderation models were observed. Mediation models—confirmed by the Sobel test—suggested STARS program targets mediated several outcomes. That is, direct instruction in self- and social-awareness and social competencies fully mediated (t = 2.407, p = 0.016) improvements in behavior at posttest. Teacher-rated relationships partially mediated (t = 1.62, p = .113) changes in student behavior. STARS was delivered with 89.82% fidelity and teachers and students indicated STARS was both feasible and socially acceptable.
Implications: Findings extend emerging research suggesting social and emotional competencies are malleable through direct instruction. That is, regardless of influences beyond the school walls, students are responsive to data-driven school-based supports. In addition, quality prevention starts with quality data. Screening 762 students is an important and feasible practice implication. STARS also provides school social workers with built-in, on-going assessment tools to gauge student responses to program elements and a way to collaborate with teachers to influence change at the student level.