Due to an increase in unverified information that reaches people, data literacy, i.e. the understanding of how to interact with, and interpret information, becomes increasingly important for informed citizenship and responsible decision-making in a democratic society. Central to data literacy is the ability to draw conclusions based on existing data, as well as to interpret and evaluate numerical relationships, the latter of which often takes its starting point in proportional reasoning, with which many students struggle. Solving proportional reasoning problems is often assessed through problems that resemble traditional mathematics tasks, which can evoke negative emotions, such as anxiety or low self-confidence in students, particularly those who already perceive mathematics as a difficult subject. This study aims to investigate if framing of problems may affect performance of different type of data literacy problems. Its further purpose is to explore potential differences between students of different self-reported difficulties with mathematics, as well as the teacher’s assessment of the student’s mathematical skills.

In this between-group study, 81 Swedish 7th graders' performance of identical proportionality problems was compared across two different conditions, namely problems that were more contextualised and therefore made less maths-like, and problems that were less contextualised, and more similar to typical maths problems found in textbooks. The questions were designed to assess proportional reasoning, as mentioned - a key component of data literacy. Each question required students to determine the truth of a statement and justify their answer by selecting one of four multiple-choice options.

The multiple linear regression analysis showed no significant difference in overall performance between students in the contextualised and non-contextualised condition. However, once the analysis was broken down to address each condition individually, teacher assessment correlated strongly with performance in the contextualised condition, but this pattern was not found in the non-contextualised condition. Instead, self-reported difficulty was the driving predictor of performance in non-contextualised condition (i.e. more maths-like problem). These findings suggests that framing data literacy problems in a broader context (non-maths-like context) may help mitigate the effects of experienced mathematics difficulty and improve performance, as students draw less associations to the subject mathematics. These results may also have implications for how data literacy is taught and studied.

Further, this study highlights the potential benefits of contextualising data literacy education to enhance student engagement and performance. Further research is needed to explore the long-term effects of such pedagogical strategies and their implications for teaching data literacy in schools.

Keywords: Education, Proportional Reasoning, contextualization, pedagogical strategies.