The Relationship Between Computational Thinking and Mathematics Understanding in Integrated Elementary Mathematics Instruction

• In Event: Assessing Computational Thinking: Updates and Advances in the Field

Abstract

Introducing computational thinking (CT) in the elementary grades has to account for what Leftwich and Yadav (2021) described as a “zero-sum game, where adding a subject means something else must be removed” (p. iv). To address this issue, CT is often integrated into core academic areas such as mathematics. However, little is known about assessing CT in integrated instruction. Beyond assessment challenges related to the ill-defined nature of CT, many questions exist about how CT should be assessed, especially when instruction is tied to other disciplinary content (Weintrop et al., 2021). For example, if students struggle with disciplinary knowledge, will those barriers influence performance on CT assessment items? (Luo, Israel, & Gane, 2022). Answering these questions is imperative to gain a better understanding of the relationship between disciplinary and CT learning. This study explored the extent to which mathematics understanding is related to CT understanding when the two subject areas are taught in an integrated classroom. The main research question was: What is the relationship between mathematical and CT understanding for 3rd and 4th grade students after participating in integrated fractions + CT instruction, and are the assessed constructs distinct from one another?

Method

Assessments
In each of 3rd and 4th grade, a series of lessons were developed aligned to CT learning trajectories (i.e., sequence, repetition, decomposition, conditional logic, variables, and debugging). These lessons were designed to fit within the Everyday Mathematics (4th edition; EM4) fractions instructional sequence and included a combination of unplugged (e.g., non-computer-based) lessons and computer-based activities that utilized the Scratch programming environment. Three CT assessments were developed for each of the 3rd and 4th grade lesson sequence, providing an early and late assessment for each grade level, resulting in four CT assessments. Our team used an evidence-centered design (ECD) process (Mislevy et al., 2003) to design our CT assessment items in alignment with CT learning trajectories (Authors, 2021b) and developed in the context of elementary mathematics, but at reduced arithmetic complexity.

Analysis
For each of the assessments, we converted the small number of polytomously scored items into binary scores based on experts’ decisions for math and CT. We conducted a parallel exploratory factor analysis (Horn, 1965; EFA) to explore the dimensionality of the response data of each assessment, thereby checking the assumptions for fitting Rasch models. We fit binary Rasch models to each of the six assessment datasets, and checked item fit, person fit, and reliability. We then used the Rasch scores to conduct an analysis to answer the research question.

Results
We found that all four CT assessments had a small relationship to their respective grade level math assessments, with only one of the correlations being statistically significant (r = .18, p < .05; the relationship between Grade 3 late CT assessment scores and Grade 3 math scores).

Conclusion and Significance
Although CT assessment items were developed in a mathematics context, there was little relationship between mathematics and CT performance. This finding illustrates that the CT constructs measured were not generally influenced by mathematics understanding.

Authors

• Corinne Huggins-Manley, University of Florida

• Maya Israel, University of Florida

• Jiehan Li, University of Florida

• Brian D. Gane, University of Kansas

• Noor Elagha, University of Illinois at Chicago

• Feiya Luo, University of Alabama

• James W. Pellegrino, University of Illinois at Chicago