Handling Missing Values in Longitudinal Panel Data With Multiple Imputation

Abstract

This article offers an applied review of key issues and methods for the analysis of longitudinal panel data in the presence of missing values. The authors consider the unique challenges associated with attrition (survey dropout), incomplete repeated measures, and unknown observations of time. Using simulated data based on 4 waves of the Marital Instability Over the Life Course Study (n = 2,034), they applied a fixed effect regression model and an event-history analysis with time-varying covariates. They then compared results for analyses with nonimputed missing data and with imputed data both in long and in wide structures. Imputation produced improved estimates in the event-history analysis but only modest improvements in the estimates and standard errors of the fixed effects analysis. Factors responsible for differences in the value of imputation are examined, and recommendations for handling missing values in panel data are presented.

Loss‐Framed Financial Incentives and Personalized Goal‐Setting to Increase Physical Activity Among Ischemic Heart Disease Patients Using Wearable Devices: The ACTIVE REWARD Randomized Trial

“…We have used this method in previous work17, 18, 19, 25 and in this study because evidence indicates that step values less than 1000 may not represent accurate data capture 26, 27. Five imputations were conducted using the mice package in R (R Foundation for Statistical Computing, Vienna, Austria), which allows for patient random effects with this data structure 37. The following predictors of missing data were included: study arm, week of study, calendar month, baseline step count, age, sex, race/ethnicity, education, marital status, household income, body mass index, days since last cardiac catheterization, most recent ejection fraction, and history of diabetes mellitus, hypertension, hyperlipidemia, smoking, and valvular heart disease.…”

The Association of Childhood Factors with Children’s Subjective Well-Being and Emotional and Behavioural Difficulties at 11 years old

“…However imputation of dependent variables is not recommended (Von Hippel 2007) and imputation of complete waves of missing data in longitudinal analysis has been found to be of doubtful benefit for regression analysis (Young and Johnson 2015). There was very little missing data within available waves for the independent variables used and so there would be no great advantage in using imputation methods for data within waves.…”

How kinship systems and welfare regimes shape leaving home: A comparative study of the United States, Germany, Taiwan, and China

“…Research demonstrates that this method yields less-biased results and more efficient estimates than complete case analysis or other traditional approaches (Young and Johnson 2015;Johnson and Young 2011). Datasets were generated by means of an imputation model, regressing incomplete covariates on the other covariates in the analysis (including interaction terms) and the outcome.…”