Data-Driven Inquiry into Algorithm Efficiency in Elementary Classrooms: A Preliminary Quasi-Experimental Evaluation

Although algorithm efficiency is central to computing practice, elementary learners rarely have opportunities to compare and improve alternative solution strategies using data. This study reports a preliminary quasi-experimental evaluation of data-driven inquiry into algorithm efficiency. Ninety sixth-grade students in four intact classes participated. Two classes received eight lessons in which students generated and refined their own solution strategies and used execution-time data across input sizes to evaluate algorithm efficiency; two active-control classes, taught by the same instructor, received procedure-focused algorithm instruction of equal duration. On the proximal, researcher-developed measure of algorithm development (0–24), the student-level pretest-adjusted difference favoring the inquiry condition was 4.90 points (95% CI [3.29, 6.52]), the unweighted class-level change contrast was 4.53 points, and the adjusted estimate was insensitive to independent-rater score replacement in a stratified 48-script subsample. Because condition was assigned nonrandomly to intact classes, these are reported as observed between-condition patterns rather than as a confirmed treatment effect. Patterns in algorithm structure understanding and self-evaluation were not robust to leave-one-class-out checks and are treated as non-robust secondary observations. The inquiry approach was implemented in regular elementary classrooms, and the pattern warrants testing with a larger number of classrooms and analyses that account for classroom-level assignment and within-class dependence.