Graphs are used in school, many occupations, and daily life, yet many people struggle to interpret them accurately. To help identify sources of difficulty in graph comprehension, we propose the Pictorial–Semantic–Task Framework. In it, we argue that accurate interpretation of graphs requires integrating pictorial variables (e.g., slope direction, graph format, data points) with semantic variables (e.g., titles, labels, scales, variable types) to determine what the graph represents. Many errors arise because readers fail to coordinate these two sources of information, often basing interpretations solely on pictorial variables. The present theoretical synthesis presents the basic analysis underlying the Pictorial–Semantic–Task Framework and an integrative review of relevant findings from graph encoding, extrapolation, and comparison tasks. The findings show that people encode and recall pictorial information far more accurately than semantic information, and often base interpretations solely on visual patterns even when semantic features call for a different conclusion. Analyses of U.S. textbooks and mass media reveal potential sources of these biased interpretations: systematic imbalances in the types of semantic information provided in textbooks and media seem likely to contribute to biases, emphasizing visual over semantic cues. By describing how perceptual and conceptual processes interact during graph comprehension, we aim to advance theories of cognitive processing in the context of graph comprehension and to derive educational implications for helping children interpret graphs more accurately.
Tsai et al. (Thu,) studied this question.