Evaluating the effect of timeline shape on visualization task performance

A depiction of all the timelines evaluated in the paper: rows divide different temporal data features (recurrent, non-recurrent and mixed) and columns divide different timeline shapes (linear horizontal, linear vertical, spiral and circle).
We evaluate the effect on task performance of 4 timeline shapes (left to right) across 3 types of temporal event sequence data (top to bottom). These images are simplified versions of the stimuli that we used in our experiment. Each dot on a timeline represents an event and has a specific categorical color to highlight where the dataset has recurrent events. Dashed lines highlight the recurrent intervals or a set of recurrent events.

Timelines are commonly represented on a horizontal line, which is not necessarily the most effective way to visualize temporal event sequences. However, few experiments have evaluated how timeline shape influences task performance. We present the design and results of a controlled experiment run on Amazon Mechanical Turk (n=192) in which we evaluate how timeline shape affects task completion time, correctness, and user preference. We tested 12 combinations of 4 shapes — horizontal line, vertical line, circle, and spiral — and 3 data types — recurrent, non-recurrent, and mixed event sequences. We found good evidence that timeline shape meaningfully affects user task completion time but not correctness and that users have a strong shape preference. Building on our results, we present design guidelines for creating effective timeline visualizations based on user task and data types. A free copy of this paper, the evaluation stimuli and data, and code are available osf.io/qr5yu/.

PDF | Preprint | DOI | Supplement | Video Preview | Video Presentation | Video Presentation (Archive) | Multiple Views Blog | BibTeX


Data Visualization @ Khoury — Northeastern University
West Village H, Room 302
440 Huntington Ave, Boston, MA 02115, USA