The Effects of Display Characteristics on the Bias of Estimates of Whisker Length of Regular and Notched Boxplots
This study investigated the effects of display characteristics on the bias of estimates of whisker length of two different types of box-and-whisker plots. To test the hypothesis that the bias means of estimates of whisker length were significantly different for the experimental and control groups, a 2.234 mixed design was utilized. Notched box-and-whisker plots and regular box-and-whisker plots were levels of the between-subjects factor, while within-subjects factors were the interquartile spread, whisker length, and spatial orientation of the graphs. Results indicate that although there was no significant difference in the bias means associated with the two types of plots, bias seems to be the product of a three-way interaction between spatial orientation, interquartile spread, and whisker/spread ratio.