Instructional Change and Student Outcomes Within Collaborative Problem-Solving Learning Environments

More than 85% of federally funded research initiatives fail to produce significant positive results (Coalition for Evidence-Based Policy, 2013). Moreover, of the less than 15% that do yield positive results, none show replicability outside of their original setting (Pogrow, 2017; Kizilcec et al., 2020). Mixed results are present in even the most promising areas of work. Decades of research show Collaborative Problem-Solving (CPS) is an effective instructional method. However, between study variations of effect sizes plagues the expansion of CPS into the classroom. The purpose of this dissertation is multifold. First, I show that research in the past decade still shows that CPS is a more effective instructional method than traditional lecture-based techniques. Second, I examine for statistical differences in the four learning methods I argue comprise CPS. Lastly, I explore the impact of six moderating variables identified as being important in previous literature to help explain heterogeneous results within the body of CPS research. The selected moderating variables include student grade, study duration, subject domain, publication year, cultural collectivism and professional development. To explore these questions, I conduct a series of meta-analyses on studies drawn from research bodies of four learning methods, Cooperative, Collaborative, Problem-based, and Inquiry-based learning. Results and conclusions of this study are consistent with previous literature. Through my meta-analysis, I found CPS to provide a large, significant effect on student achievement outcomes compared to their traditionally taught peers (t = 7.49, p < 0.001, g = 0.782); and just as previous researchers noted, significant variations between studies existed (I2 = 90.6%, Q=466.75, p < 0.001). My second analysis exploring for differences in effects between Cooperative, Collaborative, Problem-based, and Inquiry-based learning showed no significant differences. Combined with qualitative examinations of these four methods in the literature review, the argument to combine these is rational. The final two analyses examined the three identified moderating variables. The subgroup analysis examining one-time and ongoing professional development showed no significant difference between them (p = .1665). Finally, results from the analysis examining the moderating variables of student grade, subject domain, cultural collectivism, and study duration. Of these, only cultural collectivism had a significant effect on student achievement outcomes (t = 2.4219 p = .0202). The results from this dissertation show that CPS continues to have a strong positive effect on student achievement, but issues identifying high-quality studies in the field persists, limiting analyses on moderating variables. However, the confirmation of the similarity of four learning methods, future work can tap into a larger body of research for both design of new primary research as well as future distillation in the form of meta-analyses and reviews.