<![CDATA[Contextual mathematics instruction is important when pupils struggle to translate everyday situations into algebraic models. However, the role of local knowledge in supporting stages of mathematical problem solving remains underexplored. This study examined changes in year 9 pupils’ mathematical problem-solving performance across Polya’s four stages after a local salt-production ethnomathematics learning sequence in Jeneponto, Indonesia. A pre-experimental one-group pretest–posttest design involved 32 pupils in three problem-based learning sessions integrating STEAM perspectives. Two constructed-response tasks on systems of linear equations in two variables used authentic salt-production scenarios to assess problem understanding, equation formulation, solution execution, and verification. Responses were scored with an expert-reviewed stage-specific rubric and analyzed using descriptive statistics, paired-samples tests, standardized effect sizes, and normalized gain. Mean scores increased significantly at every stage and overall, with a medium overall normalized gain of 0.51. Gains were highest for devising a plan (0.66) and carrying out the plan (0.53), while looking back remained the weakest posttest stage (45.31%; gain = 0.45). Posttest performance also differed significantly between the two task items, suggesting unequal task demands. By combining an authentic coastal livelihood context with stage-level measurement, the study shows how local wisdom can function as an epistemic bridge to mathematical modeling while highlighting verification as a distinct instructional need. The study contributes a culturally grounded approach for diagnosing problem-solving processes and planning mathematics support.]]>
