Measuring the influence of map label density on perceived complexity: a user study using eye tracking

ABSTRACT

We combine eye tracking and a questionnaire-based approach to explore the influence of label density on the perceived visual complexity of maps. We design two experiments in which participants are asked to search for the names of point features on maps and to rate the map complexity and legibility for different label densities. Specifically, we conduct a highly controlled experiment in which all the map variables except the label density are held constant (the controlled experiment). Then, we conduct a second experiment following the same protocol but using real maps as visual stimuli (the real-map experiment) to verify if the results of the controlled experiment were applicable to real maps. The results of both experiments indicate a significantly positive correlation between perceived visual complexity and label density and between the response time in visual search tasks and label density. Surprisingly, we observe a significant inverse correlation between the label density and two eye movement parameters (fixation duration and fixation frequency) between the two experiments. We discuss how the variables of real maps might have affected these eye movement parameters and why the results of the two experiments are inconsistent. Our findings suggest that eye tracking parameters are not reliable indicators of map complexity. These empirical results can be helpful to future map design and map complexity investigation.

KEYWORDS:

• Map complexity
• visual complexity
• visual search
• eye movement

Acknowledgment

The authors would like to thank all the reviewers for their helpful comments and suggestions.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. The label density is calculated by dividing the number of labels by the actual display size of the maps on a monitor. The dots per inch (DPI) or pixels per inch (PPI) of the monitor are more relevant than the resolution. For example, a typical 17-inch monitor with a resolution of 1280 pixels (horizontal) × 1024 pixels (vertical) has 96.42 PPI, and its physical size is 33.72 cm × 26.97 cm (909.52 cm²). If a map with 100 labels in a resolution of 1280 × 1024 is displayed on this monitor, the label density is 100/909.52 = 0.11 per cm². A mobile phone may have similar resolution (e.g. 1280 × 720) but much higher PPI (e.g. 293.72), and its physical size is 11.07 cm × 6.23 cm (68.92 cm²). Therefore, if a 100-label map with a resolution of 1280 × 720 is displayed on this phone, the label density is 100/68.92 = 1.45 per cm². According to this definition, the label density is not related to the label distribution and cannot differentiate random, uniform, or clustered distributions (Boots & Getis, 1988).

Additional information

Funding

This research is supported by the National Natural Science Foundation of China (NSFC, Grant No. 41471382), National Key Research and Development Program of China (Grant No. 2017YFB0503602), Beijing Normal University Teaching Construction and Reform Project (Grant No. 14-07-01), and China Scholarship Council (Grant No. 201606040097).