New APSA test reliably measures student ability to pick algorithmic paradigms

Computer science students are expected to match problems with the right algorithmic design paradigm—like dynamic programming or divide-and-conquer—but measuring this skill has traditionally relied on labor-intensive free-response questions or interviews. In a new paper, researchers Dip Kiran Pradhan Newar, Michael Shindler, and Seth Poulsen introduce the Algorithmic Paradigm Selection Assessment (APSA), the first multiple-choice tool designed to efficiently and reliably test this ability. The team crafted questions that require students to identify which paradigm fits a given problem, then validated the instrument using Cronbach's alpha, achieving a score of 0.73—exceeding the standard reliability threshold of 0.7.

This development marks a significant step forward for CS education research. APSA provides a standardized, scalable way to assess student knowledge across institutions, enabling researchers to evaluate the effectiveness of various teaching interventions on paradigm selection skills. The work is published on arXiv (2606.12417) and spans topics in Computers and Society, Data Structures and Algorithms, and Human-Computer Interaction. By moving from interviews to multiple-choice formats without sacrificing reliability, the team opens the door to larger-scale studies and more targeted improvements in how algorithm design is taught.