Examining the effects of an intervention on mathematical modeling in problem solving at upper elementary grades: a cluster randomized trial study

ABSTRACTAlthough mathematical modeling during problem solving has attracted increased scholarly interest, existing quantitative work in this field has largely concentrated on secondary and tertiary education. Using a cluster-randomized-trial design, this study explored the contribution of an intervention aiming to support upper elementary students’ problem-solving modeling performance. The analytic sample of the study consisted of 50 Grade 5 and 6 classes (815 students) whose teachers volunteered to participate in the study; the classes were assigned to either an experimental (25 classes) or a control condition (25 classes), each receiving five 80-minute lessons on either modeling activities or solving routine and process problems, respectively. Student problem-solving modeling performance was measured before, right after, and two months after the culmination of the intervention. The person estimates emerging from a Rasch analysis of these data were analyzed using inferential statistics and a multi-level piecewise linear growth model. The analyses showed that students in the experimental group outperformed their counterparts in the control group both at the immediate and late test administration. Additionally, fifth graders in the experimental group outperformed sixth graders in the control group. We discuss the implications of these findings for teaching modeling during problem solving in elementary grades.KEYWORDS: Cluster randomized trial studyelementary gradesinterventionmodelingproblem solving AcknowledgmentsWe would like to thank the teachers and the students participating in this study. Without their contribution, this study would not have been possible.Disclosure statementNo potential conflict of interest was reported by the author(s).Supplementary materialSupplemental data for this article can be accessed online at https://doi.org/10.1080/10986065.2023.2270088.Notes1. This limited number of experimental studies in elementary grades should not imply a scarcity of studies on mathematical modeling in these grades. In fact, there is a rich corpus of non-experimental, qualitative studies that have explored how teaching can support elementary school students in improving their mathematical modeling competence. By adopting a longitudinal design, some of these studies have shown how mathematical modeling activities spread over several years can support the modeling competence of elementary school students in lower (e.g., English, Citation2012; English, L. D, Citation2011) or upper grades (e.g., English, Citation2022, Citation2023; English & Watson, Citation2018). Qualitative studies have also shown the potential of mathematical modeling interventions occurring within a single year to improve the mathematical modeling performance of even Gr.1 (Keisar & Peled, Citation2018) or Gr.2 (Albarracin, Citation2021) students.2. Although 998 students were invited to participate in the study (which was approved by the National Center for Educational Research and Evaluation), the analytic sample included 815 students because the parents of 90 students did not give consent to their children’s participation. Also, no complete datasets were available for 93 students because of absences during the test administrations or transfers to other schools/immigration halfway through the conduct of the study.3. Based on information obtained from the school principals, 10 classes were of low SES (5 experimental and 5 control), 29 were of medium SES (15 experimental and 14 control), and 11 were of high SES (5 experimental and 6 control).4. As O’Connor and Joffe (Citation2020) note, although there is little consensus regarding the proportion of the data set that facilitates a trustworthy estimate for inter-rater reliability purposes, double-coding 10%-25% of data units is typical. Although using a higher percentage of tests for checking for inter-rater reliability would have been optimal, due to financial constraints we opted for the lower bound of this interval.Additional informationNotes on contributorsYiannis CharalambousYiannis Charalambous is an elementary school principal. He holds a Ph.D. in Mathematics Education and a Master’s degree in Mathematics and Science Education. He has special interest in problem solving and in particular in mathematical modeling during the problem-solving process.Charalambos Y. CharalambousCharalambos Y. Charalambous is an Associate Professor in the Department of Education at the University of Cyprus, specializing in Educational Research and Evaluation. His research interests include teaching effectiveness with a particular focus on understanding and measuring the work of teaching and its effects on student learning. He is also interested in teacher initial training and teacher professional development through guided reflection on practice.