Profiling university students’ contextual numeracy through PIAAC-based tasks: Proficiency levels, gender differences, and error patterns

Abstrak

Numeracy skills are crucial in higher education; however, assessment often places greater emphasis on routine procedures than on the interpretation and evaluation of authentic quantitative information. This study examines the contextual numeracy skills of Indonesian students across various PIAAC proficiency levels, investigates gender differences, and identifies specific error patterns in each task. A descriptive-comparative cross-sectional study design was applied to 406 students (58 males and 348 females) from several state universities in Padang, Indonesia, selected through purposive sampling. Data were collected using an assessment comprising 20 items adapted from the PIAAC numeracy framework, which had been validated by experts and pilot tested. Percentage of correct responses, the Mann–Whitney U test with a biserial effect size, and item-level error analysis were used in this study. Performance peaked at Level 2 (79%) before declining at Levels 3 (61%), 4 (56%) and 5 (45%), indicating an increasing level of difficulty in higher-level numeracy tasks. Female students achieved higher scores than male students at all levels except Level 5, although significant differences were only found below Level 1 and at Levels 1, 3 and 4, with small effect sizes (0.012–0.230). Error analysis revealed difficulties in three-dimensional visualization, proportional and statistical reasoning, geometric modelling, and interpreting contextual constraints. These findings highlight challenges in advanced quantitative reasoning and provide a diagnostic framework to support authentic numeracy assessment and targeted instructional design in higher education.