Estimation of planned and unplanned missing individual scores in longitudinal designs using continuous-time state-space models.

What is it about?

Accelerated longitudinal designs (ALDs) are sampling schemes sometimes used for longitudinal research. They imply a great reduction in the economic and time cost of the study. This is achieved through a very high rate of planned data missingness. Moreover, most longitudinal studies present unexpected participant attrition leading to unplanned missing data. We conducted a large simulation study and generated various empirically plausible scenarios with missing data to: (a) evaluate the efficacy of ALDs for the study of psychological development and (b) propose a novel method (a latent change score model estimated in a continuous-time state-space framework) to estimate unobserved scores in variables that develop across many years, such as for example reading ability. We found that the group-level parameter estimates of the model were accurate, and the Kalman scores (KS) (one of the tools of these models) were able to adequately estimate individual unobserved scores. Such KS were used to impute (a) missing individual data points that were expected but unobserved due to participant attrition and (b) individual data points that were expectedly unobserved because they were outside the available age ranged observed for each case (i.e., to estimate the individual trajectories for the complete age range under study). These results have important implications for practitioners in psychology and education because they make it possible to accurately forecast individual longitudinal trajectories and to make individual-level decisions considering the model predictions. We provided the R code so that other researchers can apply this procedure to their own data. Importantly, it can be applied to a variety of scenarios where repeated measures exist for different individuals and one wants to estimate unobserved scores for specific time points (i.e., the data need not be collected through an ALD).

Featured Image

Photo by Stephen Andrews on Unsplash

Photo by Stephen Andrews on Unsplash