Evaluating User Interaction with Cartographic Visualizations on Mobile and Desktop Platforms

Not in this study. Across nine tasks, the median accuracy difference between desktop and smartphone was zero, and no task showed a significant difference after correction for multiple comparisons. Errors were driven mainly by task difficulty and by participants misunderstanding what to compare, not by the device. The caveat is that the map application was responsive and the tasks clearly defined, and participants were cartography students.

On a small screen people lose the overview, so they compensate with many short pan and zoom steps and repeated "inspect–adjust–inspect" cycles around pop-up windows. Smartphones required significantly more actions in every task (median differences of roughly 7–33 actions per task). Yet completion time was mostly comparable: these extra actions are short and cheap, so interaction count and time measure different things.

Different strategies. When each participant’s actions were converted to proportions, so that the higher totals on phones no longer mattered, phone profiles were still dominated by panning (often over half of all actions), while desktop profiles on choropleth maps leaned towards pop-up queries. The same participants shifted their mix when they switched devices, and the order in which they used the devices did not explain it.

No. For choropleth maps, where values had to be read from pop-ups or compared across layers, desktop and smartphone strategies separated clearly. For point symbol maps, both devices converged on scanning the map by panning, with only subtle differences. Task and map type therefore shape interaction at least as much as the device.

More experienced students used the legend and layer controls more purposefully, while beginners often skipped the legend after training and relied on clicking features for pop-up values. Expertise changed which tools were used rather than how often people interacted. Strategies still varied a lot within every group: some advanced students never opened the legend in point symbol tasks.

Because phone users reach the same answers through many more navigation steps, mobile map interfaces should cut unnecessary interactions: keep the legend visible or one tap away without covering the map, simplify layer management, reduce the steps needed to retrieve attribute values, and make pop-ups compact and touch-friendly.

It was a controlled lab study with cartography students from one university, so it may not reflect everyday map use or less experienced users. Eye-tracking on the two devices used different hardware and could not be analysed quantitatively; it was inspected by one researcher without independent coding. Despite counterbalancing, some learning transfer between devices cannot be fully excluded.