The Empirical Study of Trajectory-based Tasks in Pen+touch Interfaces

Jibin Yin, Hui Yang, Yang Gao, Bo ZhangAdvances 2015 · DOI

Abstract

Trajectory-based tasks are common in pen+touch interfaces. However, there is a lack of empirical knowledge regarding the use of fingers to perform trajectory-based tasks within the steering law paradigm. This study investigates the human ability to use fingers for trajectory-based tasks in pen+touch interfaces through a controlled experiment involving both dominant hand (DH) and non-dominant hand (NDH) conditions. The experiment demonstrates that when using fingers for steering tasks, the minimum effective tunnel width is approximately 60 pixels. Furthermore, no significant differences were found between DH and NDH performance regarding steering time and error rates for paths above this threshold. This work provides essential empirical knowledge for the future research and design of finger-performance applications.

Introduction

In the design of pen+touch interfaces, such as tablets, a functional distinction is often made between the high-precision input of a stylus and the comparatively "rough" input of a finger. While researchers have long utilized the Steering Law—a mathematical model predicting the time required to navigate a constrained path—to evaluate devices like mice and styluses, empirical data on direct finger input has been limited. This study addresses two critical questions for UI designers:

  1. What is the minimum "tunnel width" required for a finger to successfully navigate a path without excessive errors?
  2. Does the choice of hand (dominant vs. non-dominant) significantly impact performance in these trajectory tasks?

Methodology

The researchers conducted a controlled experiment with 15 right-handed participants (9 male, 6 female, average age 26).

Experimental Tasks

Participants performed steering tasks on an Apple iPad mini 2 using a custom application. The tasks involved navigating two distinct shapes: linear (straight tunnels) and circular (rings).

Figure 1: Experimental setup illustrating (a) linear and (b) circular steering task parameters.

The experiment tested nine different path widths, ranging from 30 to 220 pixels, with a fixed path length of 300 pixels. Performance was modeled using the standard Steering Law equation: $T = a + b(A/W)$, where $T$ is time, $A$ is path length, and $W$ is path width.

Physical Measurement

To ground the digital results, the researchers used digital calipers to measure the physical width of each participant's forefinger on both hands.

Key Results and Insights

The 60-Pixel Threshold

The most significant finding is a clear performance "elbow" at the 60-pixel width. For both steering time and error rates, performance remained relatively stable until the path width dropped below 60 pixels, at which point difficulty and error rates spiked exponentially. This suggests that 60 pixels is the minimum viable width for finger-based steering on high-resolution displays.

Hand Neutrality

While the non-dominant hand (NDH) was slightly slower than the dominant hand (DH), the study found no statistically significant difference in error rates between the two for paths above the 60-pixel limit. This suggests that the NDH is highly capable of trajectory tasks provided the UI elements are sufficiently large.

Physicality and Occlusion

The average finger width was measured at approximately 15.5 mm for the DH and 15.3 mm for the NDH.

Figure 5: Measured finger width values for participants' non-dominant and dominant hands.

This physical size creates significant occlusion on retina displays compared to a 1 mm stylus tip, explaining the necessity for wider interactive paths to maintain accuracy.

Error Rate Analysis

  • 30 pixels: Error rates were extremely high, reaching 43.2% for the DH and 81.3% for the NDH in circular tasks.
  • 60 pixels: Error rates dropped to manageable levels, approximately 7.7% for DH circular tasks.
  • >100 pixels: Error rates for both hands approached zero.

Conclusion and Design Implications

This research provides a concrete heuristic for mobile and tablet developers: trajectory-based touch paths should be maintained at a minimum width of 60 pixels to ensure a frustration-free user experience. Additionally, the findings support the use of the non-dominant hand for trajectory tasks in bimanual interfaces, provided the 60-pixel threshold is respected. Future work may explore more complex path geometries and the impact of different screen technologies on these performance benchmarks.

Summary updated

Abstract

Trajectory-based tasks are quite common in pen+touch interfaces. However, there is a lack on empirical knowledge of directly using fingers to perform trajectory-based tasks in the steering law paradigm. In this study we investigate the human ability of using fingers to perform trajectory-based tasks in pen+touch interfaces. A controlled experiment is designed and conducted, involving the DH and NDH manners. The experiment shows that when using fingers to perform steering tasks the minimum width of steering task tunnel is about 60 pixels. When using the dominant (DH) or non-dominant hand (NDH) to conduct steering tasks there is no significant difference between the two manners for the steering time and the error rate. The current work provides the empirical knowledge as basis for future research and design of finger performance applications.
BibTeX
@inproceedings{yin2015empirical,
  author = {Jibin Yin and Hui Yang and Yang Gao and Bo Zhang},
  title = {{The Empirical Study of Trajectory-based Tasks in Pen+touch Interfaces}},
  booktitle = {Advances},
  year = {2015},
  doi = {10.2991/etmhs-15.2015.190},
}
Figure