Investigation of Two Complexity Sources for Statistical Graphs.

Statistical graphs are becoming common in many fields. Modern authoring tools offer easy ways to create complex graphs. Despite multiple sources of advice, graph designers have little objective evidence on which to base design decisions to alleviate complexity. We aimed to fill a gap in the evidence of the difficulty created by two design features: adding one data series to a simple graph or adding dependent axes to a graph. We found main effects of adding a data series, of a greater number of data points, and an additional dependent axis, but primarily on the time readers studied the graph. Via deeper analysis, we explained the difficulty graph readers experienced when answering questions. That explanation focuses on the selection of points from a source graph to be part of the query. Based on our results, we recommend avenues for research on graph complexity and graph comprehension measurement.