Different graph types might differ in group comparison due to differences in underlying graph schemas. This study therefore examined whether graph schemas are based on perceptual features (i.e. each graph has a specific schema) or common invariant structure (i.e. graphs share several common schemas), and which graphic type (bar vs. dot vs. tally) is the best to compare discrete groups. Three experiments were conducted using the mixing-costs paradigm. The results showed that graph schemas are based on common invariant structure. Tally charts are quicker to compare groups than bar graphs. The processing time increased when the position difference of the compared groups was increased. The files: - Exp1_Bar_Dot.zip --> Raw data of Exp. 1 Bar vs. Dot - Exp2_Bar_Tally.zip --> Raw data of Exp. 2 Bar vs. Tally - Exp3_Dot_Tally.zip --> Raw data of Exp. 3 Dot vs. Tally - Stimuli.pdf --> The examples of the three types of data graphs: bar, dot, and tally. All the stimuli in the experiments were automatically generated. The blocks (pure graph type 1 vs. pure graph type 2 vs. mixed graph type 1 and 2) were randomised, as well as the quantities and positions of groups A, B, C in the graph. - Shiny_Experiment_Code_GraphicType_SS19.zip --> Please make sure that the experiment used the package r2d3 in css to create the tally chart and the R version 3.6.1. - StatisticalAnalyses.zip --> The aggregated data of three experiments as well as the codebook of all the variables and the statistical analyses.