A Framework of Interruptions in Distributed Team Environments

Jamie A. Rukab, Kathy A. Johnson-Throop, Jane T. Malin, Jiajie ZhangMedInfo 2004 · DOI

TL;DR

This paper introduces the Action Coding System (ACS), a novel framework for systematically studying interruptions in distributed team environments. By analyzing interruptions at the activity level, ACS provides a language to understand their impact on team performance, paving the way for better interruption management systems.

What Problem Does This Paper Address?

Interruptions are a pervasive issue in many high-stakes work environments, leading to decreased performance, increased workload, and even errors. While the negative effects of interruptions are well-documented, existing research often focuses on individual-level impacts and lacks a systematic framework for analyzing them within complex, distributed team settings. This paper addresses the need for a method to understand interruptions not just as isolated events, but as integral parts of team activities, especially in real-world operational contexts where traditional lab studies fall short.

Methods

The core contribution of this paper is the Action Coding System (ACS). ACS is a language designed to describe, represent, categorize, and analyze interruptions at the activity level. It's built upon two theoretical foundations: distributed cognition (focusing on system-level dynamics in cooperative environments with human and artificial agents) and activity theory (capturing the context of interruptions within human activities).

ACS provides a structured vocabulary to answer "who, what, where, when, and how" questions about interruptions. It breaks down an interruption instance into components like the receiver, the original task, the original action, the interrupting action and task, and the post-interrupting task and action. This allows for a chronological and detailed analysis of interruption events.

Data collection involved unobtrusive video and audio recording of two Biomedical Engineers (BMEs) at NASA's Mission Control Center during a two-hour shift handover. The data was then encoded using the ACS and analyzed using specialized software.

Datasets

The paper does not introduce a new dataset. Instead, it uses data collected from an observational study of two Biomedical Engineers (BMEs) at NASA's Mission Control Center during a two-hour shift handover. This real-world operational environment served as the "dataset" for applying and evaluating the ACS framework.

Key Insights and Ideas

  • Interruptions are deeply embedded in team activities: The ACS framework moves beyond viewing interruptions as mere disruptions to individual tasks, framing them as events within a broader activity system involving multiple agents and artifacts.
  • A standardized language is crucial: The ACS provides a much-needed common language for researchers and system designers to discuss and analyze interruptions consistently across different domains.
  • Context matters: Understanding the "who, what, where, when, and how" of interruptions within their specific activity context is key to managing them effectively.
  • Not all interruptions are equal: The study highlights that some interruptions are more disruptive than others, and the way they are handled significantly impacts workflow.
  • Potential for automation and redesign: The findings suggest that a significant portion of interruptions could be eliminated through information redesign or delegated to autonomous agents, freeing up human cognitive resources.

Key Results

During the two-hour observation period, the two BMEs experienced a total of 32 interruptions. The average duration of an interrupting task was approximately 2 minutes and 28 seconds. Interrupting tasks consumed a substantial portion of the BMEs' time: 46% for BME1 and 21% for BME2.

  • Modality: Telephone interruptions were the most time-consuming, accounting for 38% of interruption time, followed by DVIS voice loop (25%) and face-to-face (4.2%).
  • Task Resumption: In 66% of cases, the BMEs resumed their original task immediately after handling the interruption. However, 16% of the time, they failed to return to their original task within the observation period.
  • Types of Interruptions: The majority of interruptions were work-related (72%), with "determining medically related events" being the most frequent type (74%).

Limitations and Future Work

  • Limited Scope of Observation: The study involved only two participants and a single, albeit critical, two-hour observation period. This limits the generalizability of the findings to other domains or longer timeframes.
  • Contextual Nuances: The authors acknowledge that capturing the full contextual detail of interruptions from video observations can be challenging, especially when the "intended" recipient of an interruption is not the one who handles it.
  • Future Work: The authors suggest improving the ACS by capturing more detailed team and collaborative context. They also propose conducting multiple, shorter observations rather than a single extended one to improve generalizability. Further research could investigate how to best design systems and processes to eliminate, delegate, or assist humans in managing interruptions, potentially through groupware applications and intelligent tools.

Summary updated

BibTeX
@inproceedings{rukab2004framework,
  author = {Jamie A. Rukab and Kathy A. Johnson-Throop and Jane T. Malin and Jiajie Zhang},
  title = {{A Framework of Interruptions in Distributed Team Environments}},
  booktitle = {MedInfo},
  year = {2004},
  doi = {10.3233/978-1-60750-949-3-1282},
}
Figure