The Dorling map, a special type of thematic map, is a widely employed tool for visualizing geospatial statistical data. It transforms regions into circles with areas proportionate to statistical values while endeavoring to maintain regional spatial relationships. Notably, two types of Dorling maps emerge, distinguished by their consideration of regional continuity: continuous and discontinuous Dorling maps. However, as of yet, the discrepancies between these two types remain inconclusive. In this paper, we employ an eye‐tracking method to investigate the efficacy of Dorling maps in two common application scenarios, namely unpurposed browsing tasks and purposeful reading tasks. To this end, we administer tasks involving region search, attribute comparison/recognition/memory, conditional selection, relationship judgment, summary, and subjective evaluation. Subsequently, we perform a statistical analysis of the eye movement data of participants when they complete the above tasks in the continuous and discontinuous Dorling maps. The results indicate that the discontinuous Dorling maps are significantly better than the continuous ones in interpretation time for forward and reverse region search, selecting conditions, and judging adjacent relationships. Continuous Dorling maps significantly outperform discontinuous maps in terms of search efficiency during attribute comparison. Moreover, continuous maps significantly outperformed discontinuous maps in terms of cognitive supplementation or reprocessing of previous regions during conditional selection. This study can help users choose the right form of Dorling map visualization according to their needs.