4 - (SYM 87) Bridging the Idiographic-nomothetic Divide: Identifying Generalizable Emotional States

In recent years, the field of clinical psychology has begun to reckon with the ergodicity issue, the notion that statistics estimated from between-subjects data cannot be relied upon to provide within-subject inferences (Fisher et al., 2018; Molenaar, 2004). Parallel to this, there has been an upswell in the use of ambulatory measurement techniques, such as ecological momentary assessment, that allow researchers to model data idiographically (i.e., one person at a time). Nevertheless, our field strives to identify generalizable features. How then do we bridge the idiographic-nomothetic divide? That is, how can we draw generalizable conclusions about human functioning while also acknowledging nonergodicity?

The answer will likely be found in new approaches to psychological science that identify both the shared and unique attributes of psychological functioning, distilling the general structural features of emotional and behavioral systems from the broad idiosyncrasy and heterogeneity of individual persons.

The proposed talk presents one such approach. Here, we demonstrate that multivariate time series of continuously measured emotion can be discretized into a generalizable latent state structure, consisting of a baseline state, a positive emotion state, and two negative emotion states. We show how this structure replicates across three samples, a clinical sample of N=176 and two community samples of N=308 and N=738 (total N=1,222). We demonstrate how these states bridge the idiographic-nomothetic divide by explaining a substantial portion of the variance in person-level (i.e., idiographic) data on a person-by-person basis. That is, a single set of states, shared not only by the entire sample, but across all three samples, is able to explain momentary emotion variation in all 1,222 participants.

Additionally, we demonstrate how these discrete emotion states can be used to calculate the information entropy of the emotional system and how information entropy, in turn, describes the frequency and duration of valenced emotion states. Across the three samples, entropy best explained the frequency and duration of baseline emotion state frequency (R2 = 0.90, 0.77, and 0.81, respectively), followed by positive emotion state frequency (R2 = 0.64, 0.58, and 0.59), and then negative emotion state frequency (R2 = 0.58, 0.34, and 0.34). Likewise, information entropy most strongly predicted the duration of baseline emotion states, followed by positive and then negative emotion duration.