Anind Dey, Ph.D., dwellSense Principal Investigator, Carnegie Mellon University
One idea that pervades each of the Project HealthDesign studies is that computing and advanced technology has the potential to change both people’s impressions of their health and their actual health. At least that’s the promise. One of the challenges is that these positive changes often occur over a long period of time, and usually require individuals to integrate a new technology into their lives.
In human-computer interaction, the research field I call home, most of the studies of technologies for behavior change occur over a few weeks, with only a few lasting more than a month. The difficulties in running a longer study are well-documented: recruitment, attrition of participants, maintaining ecological validity while keeping the study as “clean” as possible, researcher fatigue, analysis strategies, funding, etc. Additionally, researchers are able to conduct research and get papers published without conducting such long studies.
The benefits too are well-documented: a long-term study may be the only reasonable and ecologically valid way to study the impact of a technological intervention, novelty effects of introduced technologies are minimized, there’s more time to test out technology in the field and to evolve it before needing to collect data, it gives time for users to feel comfortable with researchers and introduced technologies, and it provides an opportunity to study phenomena that occur over a long period of time.
In our project, we have the opportunity to conduct a long-term study with a few of the early-stage users of our technology. We reported on their use of visualizations of their activities after they had used our embedded assessment system for four months. Our subjects have now used our system for about a year, and we are planning on a follow-up study to understand 1) whether the problems they noticed in the performance of everyday activities caused them to change their behavior to improve performance, 2) whether repeated but infrequent reflection on their data helps them understand their performance data better than with a one-time reflection, and 3) whether our users still find value in the embedded assessment system now.
Although short-term studies of such technologies are useful and common, truly understanding how performance of activities changes over time requires a long-term study.