A recent meta-analysis performed by Sinha and Kapur utilized 53 studies from around the world comparing the order of instruction for a range of learners, primarily focused on 2nd graders through undergraduate students.  The focus of the meta-analysis was to investigate outcomes for students exposed to learning that introduces problem solving prior to instruction (PS-I) compared with students exposed to learning that introduces instruction prior to problem solving (I-PS).  Arguments in favor of PS-I designs emphasize the acquisition of higher-order thinking skills developed by allowing students to grapple with concepts they have not yet formally studied, while arguments supporting I-PS designs suggest that direct instruction is needed to enable students to focus on the most important aspects of the material.  Within the broader category of PS-I designs, the authors focused on the effects of productive failure (PF), where the problem-solving portion of the lesson is specifically designed to result in failure that can later be explored and used for student learning.

The review included 53 studies using experimental or quasi-experimental designs which performed comparisons between PF and I-PS, outcomes related to conceptual knowledge or transfer, and citing one of the seminal studies of PF or key follow-up studies of PF.

• The results of the meta-analysis indicated that PS-I designs were more effective than I-PS designs (ES = +0.36).

The authors used seven characteristics of PF to evaluate the effectiveness of this method and found higher effect sizes for PS-I designs when at least one of these characteristics was utilized.  Of the seven characteristics explored, two were found to be statistically significant moderators:

• instruction that builds on student-generated solutions (ES = +0.56 vs ES = +0.20)
• group work (ES = +0.49 vs ES = +0.19).
• While most of the effect sizes calculated in the analysis supported PS-I designs, younger students (2nd grade through 5th grade) appeared to benefit more from I-PS designs than older students.

Overall, the authors interpret the findings as largely supportive of PS-I designs while acknowledging that student age may play a significant role in determining the most appropriate and effective instructional design.

Source (Open Access): Sinha, T., & Kapur, M. (2021). When Problem Solving Followed by Instruction Works: Evidence for Productive Failure. Review of Educational Research, 00346543211019105. https://doi.org/10.3102/00346543211019105