Data Visualization Literacy as an analytical capacity in the public sector: Latent structure and response processes under time constraints

The digital transformation of the public sector has made data visualizations a key interface between analytics and decision-making. Measurement of data visualization literacy (DVL) in public administrations remains limited, and timed performance is treated as a direct proxy for competence. Such scores, however, combine two different things: whether an answer is produced in time and whether it is correct. This study addresses both gaps by examining Mini-VLAT structure and timed-score interpretation in 1,306 public employees. DVL was measured with the Mini-VLAT, a twelve-item graphical interpretation test administered under its standard 25-second-per-item limit. The analysis combines confirmatory factor analysis (CFA), testing whether the twelve items measure a single ability, with a two-part model separating response emission from accuracy and testing individual, organizational, and item correlates. Results support an essentially unidimensional DVL structure. However, response participation and interpretive accuracy show distinct correlates across the workforce: age is negatively associated with both, education mainly with accuracy, and administrative rank with both. Stacked compositional items are associated with lower performance at both stages, consistent with higher processing demands. Timed performance should not be read as a direct measure of competence. These distinctions inform training design and low–cognitive-load visualization environments for data-driven decision-making in government.