Its a tool for asking questions Often couched

It's a tool for asking questions Often couched methodologically, that self-reports cannot be trusted because people can't remember, can't select, can't aggregate, or can't report accurately what is going on. These are lesser reasons.

1. Ecological Validity – what does happen as opposed to what can happen. 2. A "Closer Relation to Life" (Lewin) 3. Taking Context Seriously (Barker, 1954) 4. Describing and Developing Taxonomies "It is impossible to describe the behavior of something when you don't know what it is. " Watson, p. 23

We cannot lose sight of the fact that we will not do away with self-reports. Social sensing data are very cool and they provide great snapshots but people's experience of their experiences is often what is most important. (Of course the comparison of social-sensing data and people's experience is particularly interesting. ) Examples from our work: -- Emotional experience and Q-T (low-grade positive emotion shorten but so do negative emotions). -- Temperament leads people to differently perceive the intensity of events. -- Loneliness people globally describe things as worse but when you look closely, their social activities are different only in certain ways (intimacy with existing close friends). -- Conflict doesn't necesarily impair relationships.

• What we need: • -- more ways to unobtrusively characterize social relations – who else is present; synch behavior records (like synchrony); get reports in real-time without destroying the ebb-and-flow; visual records of what is going on (above and beyond EAR).