Conference Agenda

Research on children’s socio-emotional development often relies on describing and classifying individual growth trajectories, a task that involves several analytical challenges - ensuring measurement stability, estimating meaningful change patterns and individual variability, and modeling nonlinear developmental trajectories. This work approaches some of these challenges, focusing on Prosocial Behavior (PB) in early childhood. PB is broadly defined as voluntary actions intended to benefit others, including sharing, helping, and comforting (Eisenberg et al., 2002). Prior evidence suggests that PB tends to increase during early childhood and is linked to later psychological adjustment (Hastings et al., 2007).

We estimated average PB trajectories through structural equation modeling and examined the consistency between the clustering solutions provided by model-based and algorithm-based methods to classify trajectories. Data included caregivers’ ratings of 2452 children (51% boys; Mage = 3.21 years, SD = 0.09), collected at four-time points, as part of the starting cohort of the National Educational Panel Study (NEPS Network, 2024), a large-scale German cohort study that follows children from birth over a 10-year period.

Latent growth curve analysis indicated an overall positive trajectory of PB over time, with a slight deceleration and an inflection point between ages 5 (T1) and 7 (T2). Additionally, results from Growth Mixture Modeling (GMM) supported a four-class solution, with classes differing in both initial levels (intercepts) and polynomial growth terms. Comparison with algorithm-based methods, such as K-means longitudinal clustering, showed moderate consistency across approaches, suggesting that each method captures unique data patterns.

Gender-equitable evaluations in teaching are a central aspect of academic equality policy and an important step towards a diverse and inclusive higher education landscape. Klonschinski (2022) shows that female teachers are systematically rated worse than their male colleagues for comparable teaching performance. Such gender-specific biases can arise from various factors. On the one hand, gender stereotypes play a role, which lead to female teachers often being evaluated based on stereotypes about "caring" and "emotional support", while male teachers are more likely to be judged on their professional competence and "entertainment factor" (Klonschinski, 2022). These different evaluation patterns contribute to the systematic disadvantage of women in teaching. In addition, recent studies indicate that the evaluation patterns in strongly male-dominated areas such as the police service may be more intensely influenced by stereotypical expectations.Our study examines whether gender biases occur in student teaching evaluations, based on a quantitative analysis of an extensive dataset from the HSPV NRW. Our study is based on data from 6002 courses between 2020 and 2024 with a total of 67,589 evaluated student feedback forms. The results contribute to the current discussion on gender inequalities in academic evaluation culture. In contrast to the effects described by Klonschinski (2022), our study shows no systematic gender bias. However, main effects of degree program, subject and teacher status as well as significant interactions are identified, indicating context-specific effects.

Transitioning to singlehood is widely recognized as a risk factor for poor well-being, but it is not yet well described how the experience spousal loss or divorce exactly affects well-being trajectories and how social resources may mitigate those effects. To obtain richer causal inferences than were previously available, we leverage synthetic control methods in analysis of eleven biennial waves (1998–2018) of longitudinal data from 7,234 participants (M_age = 66 years, 56 % women, 75% White) of the Health and Retirement Study (HRS). Specifically, we examine differences in well-being trajectories between individuals who experienced spousal loss (n = 1,084) or divorce (n = 180) and synthetically constructed non–transition control groups of continuously-married and never-married older adults with the same pre-transition well–being trajectories. Results indicated that the death of a spouse caused significant decreases in psychological well-being of the bereaved individuals in the two years prior to and four years after the event. In contrast, there was no evidence of such systematic impacts of divorce. In follow-up analyses of individual differences, we found only minimal evidence of a buffering role for social resources. Those with larger social networks experienced less steep spousal loss-related increases in depressive symptoms. Older age and non-White racial identity also mitigated the negative impacts of spousal loss. Taken together, our analyses illustrate the utility (and limits) of applying synthetic control methods to longitudinal panel data to obtain causal inferences about how major life events shape well-being during adulthood and old age.

The assumption that chronic work stressors are key drivers of burnout forms the cornerstone of many occupational health theories. Traditionally, studies investigating this relationship have relied on cross-sectional data or longitudinal analyses using the cross-lagged panel model (CLPM). In this presentation, we start with a discussion of limitations in studies aiming to understand such long-term causal processes. We show that previous approaches are often limited in their ability to inform about causal processes because they conflate within-person changes with between-person differences. Ideally, randomized controlled trials manipulating job stressors would provide clearer insights into burnout’s etiology. However, such experiments are often impractical, ethically challenging, and unlikely to yield actionable insights. Similarly, intervention designs offer limited understanding of the development of burnout, as intervention effects (e.g., offering social support) differ fundamentally from causal effects (e.g., increasing social conflict). We show how longitudinal within-person designs offer a useful tool for exploring long-term causal processes by drawing on data from 2,131 German-speaking employees over five time points spanning 24 months. We compare findings from the CLPM and the random-intercept CLPM (RI-CLPM), evaluating their respective strengths and limitations for establishing long-term causal effects. Additionally, we explore the potential of testing dynamic moderation effects through latent interactions within these models, highlighting the novel insights such analyses can yield. Finally, we examine the critical role of time in evaluating long-term effects and suggest how psychological theory can advance by addressing these temporal dynamics more explicitly.