Missing Data in Discrete Time State-Space Modeling of Ecological Momentary Assessment Data: A Monte-Carlo Study of Imputation Methods

TL;DR

This study evaluates various imputation methods for handling missing data in ecological momentary assessment time series, finding Kalman filtering most effective and challenging the efficacy of multiple imputation.

Contribution

It provides a comprehensive Monte Carlo comparison of imputation techniques for missing data in state-space models of EMA, highlighting the superior performance of Kalman filtering.

Findings

Kalman filter outperforms other methods in most missing data scenarios.

Multiple imputation struggles to accurately recover parameters in EMA data.

Missing data mechanisms influence bias and variability in estimates.

Abstract

When using ecological momentary assessment data (EMA), missing data is pervasive as participant attrition is a common issue. Thus, any EMA study must have a missing data plan. In this paper, we discuss missingness in time series analysis and the appropriate way to handle missing data when the data is modeled as an idiographic discrete time continuous measure state-space model. We found that Missing Completely At Random, Missing At Random, and Time-dependent Missing At Random data have less bias and variability than Autoregressive Time-dependent Missing At Random and Missing Not At Random. The Kalman filter excelled at handling missing data under most conditions. Contrary to the literature, we found that, using a variety of methods, multiple imputation struggled to recover the parameters.