The role of data visualization literacy in public administration in the era of AI adoption

What the paper says

As Artificial Intelligence (AI) systems and data-driven tools become integral to governmental decision-making, the ability to interpret and reason with visual information emerges as a critical competence for operating effectively in AI-mediated analytical environments. However, empirical evidence on the level of data visualization literacy within public administrations remains limited. To address this gap, the study provides a large-scale, diagnostic, and descriptive analysis of Data Visualization Literacy (DVL) performance in a real public organizational setting, using a standardized assessment instrument. A cross-sectional survey of 1,219 public employees was conducted using a bilingual Spanish–Valencian adaptation of the Mini-VLAT (12 items; 25 seconds per item), evaluating participants’ capacity to interpret, analyze, and reason with graphical representations of data. Mean performance reached 57.8% correct, with 27.1% omissions and 15.1% errors. Tasks involving proportional or relational reasoning—particularly stacked charts—produced the lowest accuracy and the highest nonresponse. Performance patterns were consistent: accuracy declined with age, improved with higher educational attainment, and varied across departments. Omissions under time pressure, rather than misinterpretation, were the predominant source of error. The findings underscore the importance of treating DVL as part of the institutional infrastructure, through periodic diagnostics, shared graphic-interoperability standards, targeted domain training under time constraints, and longitudinal monitoring to preserve epistemic control while harnessing AI’s speed and scale.