Latent State-Trait Modeling of Satisfaction with Life Scale: An Item-Level Analysis Using Dutch Panel Data

This study sought to examine the stability and change of the 5 items of the Satisfaction With Life Scale over several years. The multi-trait-multi-state model was used to separate the variance components attributable to stable influences, time-varying situational influences, and random measurement error. The study used two partially overlapping samples of Dutch adults surveyed annually over 8 and 5 years (N1 = 9,300, N2 = 6,770). Results show that between 54% and 63% of the variance in items is stable over time, whereas between 6% and 26% of the variance is due to occasion-specific influences (i.e., situational influences present at each measurement time and person-situation interactions). Between 15% and 36% of the variance is due to measurement error. All items show a similar degree of stability over time. However, the first 3 items (measuring current life satisfaction) show higher reliability, whereas the last 2 items (measuring past life satisfaction) are less reliable. The study has several implications for measuring life satisfaction in cross-sectional and longitudinal studies. For example, the results suggest that the first 3 items are more sensitive to and better reflect current situational factors than the last 2 items. In contrast, the last 2 items reflect more unsystematic and random influences that are more independent of current life situations. Thus, relying on the last 2 items may lead to an underestimation of the reactivity of life satisfaction judgments to current life situations.