Biasing and debiasing health decisions with bar graphs: Costs and benefits of graph literacy

Abstract

Bar graphs can improve risk communication in medicine and health. Unfortunately, recent research has revealed that bar graphs are associated with a robust bias that can lead to systematic judgment and decision making errors. When people view bar graphs representing means they tend to believe that data points located within bars are more likely to be part of the underlying distributions than equidistant points outside bars. In three experiments we investigated potential consequences, key cognitive mechanisms, and generalizability of the within-the-bar bias in the medical domain. We also investigated the effectiveness of different interventions to reduce the effect of this bias and protect people from errors. Results revealed that the within-the-bar bias systematically affected participants' judgments and decisions concerning treatments for controlling blood glucose, as well as their interpretations of ecological graphs designed to guide health policy decisions. Interestingly, individuals with higher graph literacy showed the largest biases. However, the use of dot plots to replace bars improved the accuracy of interpretations. Perceptual mechanisms underlying the within-thebar bias and prescriptive implications for graph design are discussed.

Promoting the use of probabilistic weather forecasts through a dialogue between scientists, developers and end‐users

“…When reading barplots, people tend to believe that points within the bar are more likely than outside the bar – even if they are equidistant relative to the top of the bar (Newman and Scholl, ; Okan et al . ). FeWIS Pro's barplots (Figure d) factor this tendency in by displaying the 90%ile as the full height of the stacked bars, which implies that values outside the bar are in fact less likely than those within the bar.…”

Measuring Graph Literacy without a Test

“…23 More graphliterate individuals also benefit more from interventions designed to improve risk understanding or to reduce errors and biases in graph comprehension (e.g., benefit from error bars). 21,27 Despite its success, like all scales, the OGL scale has some limitations. One of the more pressing issues is that it typically requires 9 to 17 minutes to complete, 10,16,25 which may be too long for some uses (e.g., in clinical practice).…”

Meeting the challenge of the Psychonomic Society’s 2012 Guidelines on Statistical Issues: Some success and some room for improvement

“…Newman and Scholl (2012) demonstrated that the use of bar charts to present means leads to a within-the-bar bias such that values within the bar are perceived as more likely than values outside (e.g., above) the bar (see also Fritz, Morris, Cherchar, Smith, & Roe, 2015;Okan, Garcia-Retamero, Cokely, & Maldonado, 2017).…”