Modelling non-linear personality change surrounding transitions: A review of statistical approaches

Personality changes surrounding transitions in life circumstances are often non-linear, presenting challenges for statistical analysis. This paper therefore reviews approaches to modelling non-linear personality change surrounding transitions, aiming to guide readers in selecting and applying an approach that fits their objectives. Seven approaches were reviewed, including traditional mixed-effects methods, continuous-time dynamic models, and relatively novel data-driven techniques. Each approach is explained, outlining its strengths and limitations. The approaches' practical utility is assessed through a case study examining changes in life satisfaction surrounding widowhood, using LISS panel data. Interpretability and model fit are compared, and annotated R code is provided as a tutorial for implementation. Results highlighted the varied suitability of the mixed-effects approaches for studying different aspects of change. The data-driven techniques excelled in capturing average and person-specific trajectories, generalised effectively, and allowed interpretation of different change aspects than the mixed-effects approaches allowed for. Importantly, the approaches yielded distinct findings regarding life satisfaction changes surrounding widowhood, with theoretical implications. The paper concludes with practical recommendations for selecting and applying these approaches. By expanding the reader's statistical toolkit and providing an accessible overview, this resource supports the effective analysis of non-linear changes surrounding transitions, enabling a fuller understanding of personality change.