Methods for Handling Missing Data in Resilience Research Using Latent Trajectory Studies: A Systematic Review of Current Practices and Best Practice Recommendations

This systematic review aims to provide a comprehensive overview of current practices in handling missing data in resilience research when using longitudinal mixture modeling, specifically growth mixture modeling (GMM), latent class growth analysis (LCGA) or group-based trajectory modeling (GBTM). The project leverages the ResiMeta database, which contains over 300 longitudinal studies examining resilience processes, to systematically identify and evaluate the methods researchers currently employ when confronted with missing data in trajectory-based analyses. Resilience research is particularly vulnerable to missing data challenges, as stressor exposure and deteriorating mental health are systematically associated with study dropout, creating a critical methodological concern that can substantially bias findings about prevalence rates of resilient trajectories and the predictive value of resilience factors. The review follows a two-pronged approach. First, we will systematically extract information from empirical studies included in the ResiMeta database to document which missing data methods are actually used in practice, how researchers report their missing data mechanisms, and whether they conduct sensitivity analyses to assess the robustness of their findings. This will provide an evidence-based overview of the current state of the field, enabling the identification of potential gaps between methodological best practices and actual implementation. Second, we will conduct a parallel systematic literature review of methodological papers and simulation studies that provide recommendations for handling missing data in longitudinal mixture modeling. These papers will be systematically evaluated to identify evidence-based best practice recommendations regarding when to use specific methods such as full information maximum likelihood (FIML), multiple imputation (MI), or other approaches, and under which conditions different methods perform optimally. By comparing the actual practices documented in empirical resilience studies with the methodological recommendations derived from simulation studies and methodological papers, we aim to identify discrepancies between what researchers do and what methodological evidence suggests they should do. This comparison will allow us to develop concrete, field-specific best practice recommendations that are both methodologically sound and practically feasible for resilience researchers.