Amelia C. Warden¹ (), Jessica K. Witt²; ¹University of Michigan, ²Colorado State University

Effectively conveying time-series data is critical for decision-making in many domains, like climate science, economic forecasting, and epidemiology. Oftentimes, designers opt to use colors that are semantically compatible with the underlying data, such as red for higher and blue for cooler temperatures. An alternative method is to use colors that better exploit ensemble processes, which refer to the visual system's innate ability to extract summary statistics, like the mean, from a set of similar objects. Our prior work examining trend detection in visualizations found that single-hue color palettes, which engage visual system processes, better convey temperature trends than semantically compatible color palettes. While assessing how color-coded visualizations impact trend detection in a politically and emotionally significant context is important, this context carries inherent biases that may influence perceptions and interpretations. To further explore the efficacy of color palettes that engage the visual system, the current work examines trend detection for data representing a more impartial context with fewer inherent biases, specifically red and blue car sales over time. Participants viewed stripplot graphs and indicated whether trends were increasing when the underlying data was encoded with either a single-hue color palette exploiting ensemble processes or a semantically intuitive multi-hue color palette. Using signal detection theory, the results revealed significantly higher sensitivity (d’) to trends presented with a single-hue color palette than a more semantically compatible multi-hue color palette. These findings highlight the generalizability of using color palettes that better engage properties of the visual system, suggesting they improve trend detection independent of the underlying context of the data. Additionally, these findings further demonstrate the advantages of single-hue displays in enhancing ensemble perception. The results have implications for broader applications of ensemble-based color schemes used in information visualizations, which can improve public comprehension and decision-making when adopted for time-series data.