Evaluation of Noise Annotation Lines: Using Noise to Represent Thematic Uncertainty in Maps

Noise annotation lines are a promising technique to visualize thematic uncertainty in maps. However, their potential has not yet been evaluated in user studies. In two experiments we assessed the usability of this technique with respect to a different number of uncertainty levels as well as the influence of two design aspects of noise annotation lines: the grain and the width of the noise grid. We conducted a web-based study utilizing a qualitative comparison of two areas in 150 different maps. We recruited participants from Amazon Mechanical Turk with the majority being nonexperts with respect to the use of maps. Our findings suggest that for qualitative comparisons of ‘constant uncertainty’ (i.e., constant uncertainty per area) in thematic maps, noise annotation lines can be used for up to 6 uncertainty levels. During comparison of 4, 6, and 8 levels, the different designs of the technique yielded significantly different accuracies. We propose to use the ‘large noise width, coarse grain’ design that was most successful. For ‘mixed uncertainty’ (i.e., uncertainty is not constant per area) we observed a significant decrease in accuracy, but for 4 levels the technique can still be recommended. This article is a follow-up to our conference paper reporting on preliminary results of the first of the two experiments (Kinkeldey et al. 2013).