Shannon Dickson

Shannon Dickson

Quantitative Psychology and Individual Differences
Faculty of Psychology and Educational Sciences
KU Leuven

Email
Website

Project
Measuring the dynamic structure of affect

Many emotion researchers study the dynamics within and between momentary positive affect (PA) and negative affect (NA) across time, by means of the Experience Sampling Methodology (ESM): Using a smartphone app, participants respond to a number of questions (items) asking about the presence and intensity of positive and negative emotions at random moments throughout the day for the duration of one or more weeks. Based on the reported scores on multiple specific emotions, researchers then compute a PA and NA score. This is often done by summing or averaging the scores on positive and negative emotions, respectively. Next, they model the dynamics of the PA and NA scores by means of the vector autoregressive model (VAR). In a VAR model, momentary PA and NA are predicted based on the preceding PA and NA scores. The computation of PA and NA implicitly relies on a so-called measurement model (MM) that specifies which observed items measure which construct of interest (e.g., negative emotions measure NA) and to what extent (e.g., each emotion contributes equally). In ESM studies, it is likely that the implicitly assumed MM does not hold, and such misspecification may lead to incorrect VAR parameter estimates. It is thus essential not to implicitly assume but to explicitly study the MM when fitting VAR. The existing state-of-the-art methods for doing so have the important limitation that they require the user to prespecify which items measure which construct. This is often difficult in ESM as, for instance, even the typically assumed PA-NA distinction is not systematically found. Thus, an exploratory method is needed to infer the MM from the data, that is, to figure out which items are measuring which construct. Another challenge pertains to potential differences in the MM when comparing VAR parameters across persons and over time. To ensure comparability, the MM should be (at least partially) the same across participants and time points. This is referred to as measurement invariance. ESM is prone to violations of this condition due to person- or context-specific item interpretations, and these violations should be accounted for in the analysis. Therefore, in this PhD-project, we will develop new exploratory methods to (1) determine the appropriate MM for a particular ESM data set and account for it in VAR; and (2) to account for differences in the MM across time and/or across persons.

See also KU Leuven.

Supervisors
Prof. Dr. Kim De Roover
Prof. Dr. Eva Ceulemans

Financed by
Bijzonder Onderzoekfonds

Period
1 October 2024 – 1 October 2028