CAES   Context-Aware Experience Sampling
What's new?

Design & Code

Context-Aware Experience Sampling (CAES) is the term we use for a type of data collection method based on the experience sampling method (ESM). ESM is also known as ecological momentary assessment (EMA). For an overview of context-aware experience sampling, see this short paper.

ESM [1] and EMA [3] use monitoring devices to assess phenomena at at the time that they occur as the people being observed are in natural settings. Fundamentally, the goal is to use tools that maximize the validity of data collected by avoiding or minimizing retrospective recall. ESM/EMA are forms of naturalistic observation [3]. Traditional ESM/EMA methods have four characteristics [2]. 

  • Assess phenomena at the moment they occur
  • Usually involve a substantial number of repeated observations
  • Made in the environments that subjects typically inhabit 
  • Dependent upon careful timing of assessments

Context-aware experience sampling improves upon ESM/EMA by using emerging computational perception and sensing technologies to automatically detect events that can trigger sampling and thereby data collection.

For instance, in the standard methods researchers are typically limited to three types of scheduling formats [4]: event-contingent, time-contingent, and signal-contingent. Event contingent sampling has previously required the subject to select an interaction time based upon his or her behavior (i.e. the researcher tells the subject, "when event X occurs, record your observations about Y"). Time-contingent sampling consists of simply sampling at predetermined times of the day or week. Signal-contingent sampling is when the device the subject carries can provide an investigator-controlled cue (e.g. beep) indicating that data should be collected and where the cue can be delivered with some pre-determined probability randomly within a given time frame. 

Context-aware experience sampling permits much more sophisticated event-contingent sampling where the computer detects the context or a specific event and provides the signaling cue. As awareness recognition algorithms improve, the potential exists for investigators to identify specific activities of interest (e.g. climbing stairs) and have the CAES tool sample just before, during, or after the desired activity. This may permit more extensive sampling about the situation of interest than is possible today without unduly burdening the subject.  In addition to new sampling options for investigators, the same sensors that are used for context and activity detection can be used to collect raw data itself. For instance, heart rate or positional data (i.e. GPS-determined location) may be used both as a measurement stored for future analysis and as a signal that is processed in real-time to detect if the subject is engaged in an activity of interest. 

ESM/EMA tools are most valuable when they can be used by subjects for long periods of time without supervision. Our original context-aware experience sampling software (CAES) was developed to permit this type of study and was used at MIT and elsewhere, primarily for research in ubiquitous computing. That version was for PDA devices. The code is still available to anyone who wants it, but it is not supported. Recently we have moved to using mobile phone devices. After buidling many one-off custom programs for internal use, we have recently joined forces with the MyExperience project and will be contributing to that effort.

We are using MyExperience in a study to investigate the feasibility of what we call the National Experience Sampling Project. We are also redeveloping our wireless accelerometers and automatic activity detection software to work with mobile phones for longitudinal detection of physical activity type, intensity, duration, and location. We call these sensors Wockets. The Wockets project, like the MyExperience project, is open source. Please participate!

References cited above

[1] Csikszentmihalyhi, M., Larson, R. Validity and Reliability of the Experience-Sampling Method. Journal of Nervous and Mental Disease, 1987, 175:526-536.

[2] Stone, A.A., Shiffman, S. Ecological Momentary Assessment (EMA) in Behavioral Medicine. Annals of Behavioral Medicine, 1994, 16(3): 199-202. 

[3] Barker, R.G, Wright, H.F. One Boy's Day. New York: Harper & Row, 1951.

[4] Wheeler, L., Reis, H.T. Self-Recording of Everyday Life Events: Origins, Types, and Uses. Journal of Personality. 1991, 59: 339-354.

Comments   2008, MIT