Project
Capturing Relations among Psychological Constructs and Comparing Them Within Groups

Many important research questions in psychology seek to uncover relations among psychological constructs that are not directly observable or ‘latent’ – that is, they can only be measured indirectly via questionnaires. For example, does social pressure to be happy increase or decrease wellbeing? And how does experiencing negative affect relate to satisfaction with life? Structural equation modeling (SEM) is the state-of-the-art to model these relations. When comparing such ‘latent relations’ across many groups (e.g., countries in a cross-national study), they likely differ among the groups (e.g., they may be culturally dependent) or within the groups (e.g., they may differ across subsets of a nation’s inhabitants determined by a combination of demographics that are unknown or unmeasured). This PhD-project aims to develop novel mixture SEM methods for finding clusters of subjects within groups that have the same latent relations. In this way, one can study how the groups differ in terms of these relations and how they differ within the group. For instance, it may be that, in one country, all inhabitants are homogenous in terms of what determines life satisfaction, whereas, in another country, the inhabitants strongly differ in this respect. For finding such clusters, differences in how the psychological constructs are measured by questionnaire items – for example, differences in item interpretation or translation – should be unraveled from the differences in the psychological relations of interest. An important challenge is that these ‘measurement differences’ may exist between the groups even when the differences of interest exist within the groups, or vice versa. This project focuses on mixture SEM methods that tackle such challenges by optimizing the comparability of the psychological constructs’ relations both within and between groups while, at the same time, capturing clusters of subjects within the groups with equivalent relations.

Supervisors
Prof. Dr. Kim De Roover
Prof. Dr. Jeroen Vermunt

Financed by
ERC

Period
1 October 2024 – 1 October 2028