From Quantified Self to Quantified Crowd: Studying Human Behavior in the Workplace

The revolution in sensor technology development is changing how human behavior can be studied. It enables detailed measurements of people in their natural environments–providing big data. While some forms of big data collection may be well known to the general public (e.g. mining Twitter feeds), another paradigm for big data collection is beginning to occur: measuring human behavior in everyday life.
Prof. Gloria MarkOver the last several years, with students and colleagues, Informatics professor Gloria Mark  has extended the notion of “life-logging” or “the quantified self” to the idea of measuring “the quantified crowd.” Rather than bringing people into the laboratory to study their interaction with computers, sensors enable researchers to create a “living laboratory” where people can be studied using computers in their natural environments. To study human behavior, Mark and colleagues use a range of different types of sensors for tracking stress, mood and computer behavior (e.g. heart rate variability as a measure of stress; actigraph data as a measure of activity), experience sampling for probing mood throughout the day, wearable cameras for tracking face-to-face interaction, and logging computer and phone behavior. Using this methodology, they can track people in situ, to capture how stress, attention focus and well-being change as people use information technology and conduct work in their daily environments. The goal of this data collection is to do a deep examination on the relationship of behavior (e.g. multitasking or email usage) with other factors such as attention focus, stress, and wellbeing. All data is measured to the second, time-stamped, and then synched together. The collection, processing, and analysis of the data produce “small” big data sets and present significant challenges.
This research has been conducted in two domains: the workplace and campus life. Partnering with Microsoft Research, Mark and colleagues have collected data on over 72 individuals tracked in the workplace for 1-2 workweeks. Earlier research was conducted in partnership with the U.S. Army Natick Soldier Systems Center in Natick, MA, where, in collaboration with Stephen Voida, data was collected to study the effects of cutting off email in the workplace. Funded by an NSF grant, Mark is also studying information technology use of college students, research being conducted with UCI Informatics graduate student Yiran Wang (G. Mark, advisor) and UCI Education graduate student Melissa Niiya, and in collaboration with professors Mark Warschauer and Stephanie Reich in the UCI Department of Education. To date, 124 college students were tracked in their everyday campus life, for seven days, all waking hours.
Measuring information technology use 'in situ' with different sensorsCollecting this data enables a number of questions about information technology use, stress and well-being in the workplace to be answered. For example, it was found that people exhibit rhythms of attention focus in the workplace. People need time to “ramp up” before they can be highly focused at work: on average focused attention peaks late morning and mid-afternoon whereas boredom peaks early afternoon. In terms of workplace happiness, surprisingly people are happiest doing rote work, as opposed to focused work. This could be explained by the finding that focused work is associated with stress. It was also found that when people are engaged in their work, online interactions make people happier than being in face-to-face interactions. When highly engaged in work, with online interactions, people can control their social break time. In another exploration, distractions were examined when working online. A common assumption is that one is focused in an online activity and then becomes distracted by email or Facebook, or something else. Looking at this from a reverse perspective, the researchers asked whether one might first be in a particular attentional state that makes one susceptible to distractions. Devices used to measure behavior and mood. Clockwise from upper left: active badges, experience sampling, computer logging, heart rate monitor, wearable galvanic skin response sensor, and SenseCam (wearable camera).The data revealed that this was indeed the case. When people are conducting work that is considered “rote” or boring, then people were more likely to subsequently engage in “lightweight” communications that might be considered distracting, e.g. Facebook use.
With college students, among other findings, the researchers found that the more time spent online, the more stress students experienced. However, the more time spent on social media seems to mitigate stress. Heavy multitaskers use significantly more social media and experience lower positive mood than light multitaskers. Similarly, the more constantly that a student checks social media, the lower is their positive mood. College students have a much shorter focus of attention on any computer screen on average than information
workers in the workplace.
This work has been presented at the South X Southwest Festival (SXSW) and the top conferences in the field of human-computer interaction, including ACM CHI (human factors in computing) conference and ACM CSCW (Computer- Supported Cooperative Work: The conference on Social Computing). It has garnered multiple paper awards and has received significant media attention. Mark has been interviewed by NPR and BBC news. This work has been covered in the NY Times, The Atlantic, Wall St. Journal, and other media.
This comprehensive capture of data in the real world using sensors has a number of implications for workplace and technology design. In the workplace, managers can use this data to design schedules that optimize when people are most focused. Work breaks can also be strategically designed to counteract boredom. Information technology can be designed to inhibit the switching of computer activity (and hence the rapid switching of attention). Longer periods of focus can be expected to improve productivity and well-being. The results in fact suggest this. Analyses are ongoing.
This article appeared in ISR Connector issue: