Fig 1.
Examples of poor graph design.
Examples of a poorly designed bar graph (left), pictorial area chart (center), and pie chart (right). All graphs presented in this paper were created for this study with fictional data using Adobe Illustrator 2023 and 2024.
Fig 2.
Series of bar graphs (e.g., graph 1), pictorial area charts (e.g., 5), and pie charts (e.g., 9) per stage (see far left). The graphs were randomized within each series per participant. The series consisted of accurate graphs (with a green check mark), misleading graphs (with a red cross), and corrected versions of the misleading graphs (marked with a cross and a check mark). The treatment consisted of a clean or full-design correction. Participants were randomly assigned to a group and a correction design.
Fig 3.
Examples of an accurate bar graph (left), pictorial area graph (center), and pie chart (right).
Fig 4.
Examples of misleading graphs.
Example of a misleading bar graph correction (top) in full-design, and a misleading pictorial area graph (center) and pie chart (bottom) correction in clean design.
Fig 5.
Visualizations of the hypotheses.
a. For H1 we compared evaluations of misleading and accurate graphs at baseline; b. For H2 we compared misleading graphs at baseline and accurate graphs at treatment. For H3 we included the full/clean design condition in this comparison; c. For H4 we compared new misleading and accurate graphs (stage new). For H5 we included the full/clean design condition in this comparison.
Table 1.
Demographics and graph literacy scores (GLS) of the interviewed participants.
Table 2.
Demographic information of participants, including their graph literacy sum scores.
Fig 6.
Density plots showing the distribution of the evaluation scores of the accurate and misleading graphs on the VAS scale at baseline, separate for each context (left) or together for each graph type (right).
Density plots are smoothed versions of histograms that represent the raw distribution of the answers of the participants in the various conditions. Smoothed tails are cut off at the answer range endpoints 0 and 100 and the dots indicate the mean values per condition.
Fig 7.
Density plots showing the distribution of the evaluation scores on the VAS scale, comparing the misleading graphs at baseline and their corrected version (in either clean design or full-design).
Density plots are smoothed versions of histograms that represent the raw distribution of the answers of the participants in the various conditions. Smoothed tails are cut off at the answer range endpoints 0 and 100 and the dots indicate the mean values per condition.
Fig 8.
Density plots showing the distribution of the evaluation scores of the new accurate and misleading graphs on the VAS scale after seeing corrections, separate for each context (left) or together for each graph type (right).
Density plots are smoothed versions of histograms that represent the raw distribution of the answers of the participants in the various conditions. Smoothed tails are cut off at the answer range endpoints 0 and 100 and the dots indicate the mean values per condition.