3 - (SYM 28) An Examination of Individual Networks in Context: “good” versus “bad” Days

Background: Clinical science has historically been concerned with aggregate outcomes, such as in randomized controlled trials (e.g., A-Tjak et al., 2015; Olatunji et al., 2013). However, aggregate outcomes generally fail to consider variation between and within individuals. To ensure findings are sensitive to the individual, we need idiographic (person-level) approaches to adequately represent the complexity and variation in psychopathology experienced by each participant. Although many researchers have examined person-level dynamics (e.g., Fisher et al., 2017; Kaurin et al., 2022), accounting for differences between individuals, they did not account for changes within the individual; that is, that relationships among symptoms may be different for the same person in different contexts. The current study aimed to bridge this gap by examining contextual differences in individual networks.

Method: We used ecological momentary assessment to measure psychological symptoms and functioning over 60–90 days (current N = 21, data collection in progress). At the end of each day, participants were asked to rate how their day went, from “very good” to “very bad.” Group Iterative Multiple Model Estimation (GIMME) analyses were used to identify individual-level networks and subgroups in the sample, to evaluate similarity between networks.

Results: In our sample, three participants’ “good” and “bad” day networks were clustered in the same subgroup, indicating some stability between these contexts for these participants. The other 18 participants’ “good” and “bad” day networks were assigned to different subgroups, suggesting contextual differences in individual networks. In total, we identified six subgroups with two to 10 networks per group.  

Conclusion: These findings suggest that individual-level networks measured across different contexts may not fully capture contextual shifts, weakening their reliability. For future research wherein reliability of networks is crucial, it may be necessary to consider how individuals respond in various contexts and estimate separate networks for contexts of interest (e.g., used therapy skills vs. did not use therapy skills).