Affective structure, measurement invariance, and reliability across different experience sampling protocols

Abstract

While measures of affect are often included in studies using experience sampling methodology (ESM), the affective structure at the between-and within-person level has not been thoroughly investigated. Additionally, it is currently unclear to what extent this structure is influenced by variations in sampling protocols. Our first aim was to investigate the structure of affect at the between-and within-person level when measured with 8 or 18 affect items.Second, we investigated the invariance of this structure across different sampling frequencies and questionnaire lengths. Further, the reliability of positive and negative affect (PA and NA, respectively) was calculated. Participants (N = 163) were randomly assigned to receive either a 30 or 60 item questionnaire three, six, or nine times per day over 14 days. Momentary affect was assessed with 8 or 18 items. At both levels, a two-factor structure with correlated PA and NA showed the best fit compared to an orthogonal and a unidimensional model. However, the CFAs and additional EFAs indicated that items shared additional variance that could be accounted for by freeing residual correlations or adding more nuanced affect factors to achieve acceptable fit. A structure accounting for these additional relationships was invariant across different sampling protocols at the within-person level and resulted in high reliability.We observed differences between levels in the strength of the loadings, the correlation between PA and NA, and the number of factors suggested in EFAs. Taken together, these structural differences indicate a more discrete affective structure within than between persons.

Affect-specific empathy: Experience sampling and multilevel structural equation modeling provide a within-person perspective.

“…In 87% of the time participants empathized with only one affect at a time (Depow et al, 2021). Furthermore, although the magnitude of correlation was only weak within our study, our finding is broadly in line with the results of several studies employing latent modeling, showing that positive and negative affect are moderately to highly negatively correlated at the within-person level (Brose et al, 2015; Eisele et al, 2021; Rush & Hofer, 2014).…”

“…For the within-person level, this means that individuals who felt positive affect were less likely to feel negative affect at the same time. At the between-person level, on the other hand, individuals' overall level of positivity was unrelated to their overall level of negativity, or at least individuals were slightly less likely to experience positive and negative feelings at the same time (Brose et al, 2015; Eisele et al, 2021; Rush & Hofer, 2014). For some people the within-person structure was more similar to the between-person structure relative to the within-person structure for other people.…”

“…2Removing participants from MSEM analyses due to lack of variance in their responses on the variables of interest is standard practice in the field (e.g., see Eisele et al, 2021). The interpretation of the results did not change by removing participants without any variance.…”

Emotional anticipation of future vocational transitions: A person-centered approach

“…However, future research is warranted to investigate the temporal dynamics of emotional anticipation. While longitudinal designs are good candidates for such objectives, we further argue that experiences sampling studies and intensive longitudinal studies using momentary and intense assessments are best suited given the malleable, transactional, and multi-componential nature of emotions (Eisele et al, 2021). Additionally, given the existence of both trait and state components of emotions, studies that rely on recent developments in longitudinal analysis such as random-intercept cross-lagged panel modeling (Hamaker et al, 2015;Mulder & Hamaker, 2020), and their extension to latent profile analysis (RI-LTA; Muthén & Asparouhov, 2020) could be highly informative as well.…”

“…Future research could thus better distinguish between future-oriented emotions, mood, and dispositional future-oriented affect in including and controlling for additional constructs. An additional avenue relates to the use of experience sampling studies that are well suited to disentangle the stable and state dynamics of emotions (Eisele et al, 2021;Fisher & To, 2012). This kind of approach is well suited to understand the enactment and dynamics of emotional anticipation within a day or week, which may be particularly relevant to develop a better understanding of the emotional anticipation pattern.…”