Handling Attrition in Longitudinal Studies: The Case for Refreshment Samples

Abstract

Panel studies typically suffer from attrition, which reduces sample size and can result in biased inferences. It is impossible to know whether or not the attrition causes bias from the observed panel data alone. Refreshment samples-new, randomly sampled respondents given the questionnaire at the same time as a subsequent wave of the panel-offer information that can be used to diagnose and adjust for bias due to attrition. We review and bolster the case for the use of refreshment samples in panel studies. We include examples of both a fully Bayesian approach for analyzing the concatenated panel and refreshment data, and a multiple imputation approach for analyzing only the original panel. For the latter, we document a positive bias in the usual multiple imputation variance estimator. We present models appropriate for three waves and two refreshment samples, including nonterminal attrition. We illustrate the three-wave analysis using the 2007-2008 Associated Press-Yahoo! News Election Poll.

Retail Payment Innovations and Cash usage: Accounting for Attrition by using Refreshment Samples

“…Proposition 1 rules out interactions between z 1 , z 2 and z 3 in the three-period attrition function, but it is possible that the latter is overidentified. For a binary outcome Y , Deng et al (2013) demonstrated just-identification for the joint distribution of .Y 1 , Y 2 , Y 3 , S 12 , S 23 / by counting the number of cells in the contingency table and the number of parameters in the saturated model. However, it is not obvious how to obtain the just-identification of the attrition function in the case where the outcome variable Y is continuous.…”

A Biopsychosocial and Long Term Perspective on Child Behavioral Problems : Impact of Risk and Resilience

“…Previous studies have noted similar magnitudes of attrition in longitudinal studies (Deng et al, 2013). However, selective attrition might be problematic for the generalizability of the results.…”

Bayesian latent pattern mixture models for handling attrition in panel studies with refreshment samples

“…Model Diagnostics. To check the fit of the models, we follow the advice in Deng et al (2013) and use posterior predictive checks (Meng, 1994b;Gelman et al, 2005;He, Zaslavsky and Landrum, 2010 and Reiter, 2010). We use the BLPM model to generate T 0 = 500 data sets with no missing data in (X, Y 1 , Y 2 , W ), randomly sampling from the T =1000 available completed datasets.…”

“…In previous research using the APYN data (Pasek et al, 2009;Henderson and Hillygus, 2011;Iyengar, Sood and Lelkes, 2012;Henderson, Hillygus and Tompson, 2013), scholars have mostly relied on post-stratification weights to correct for potential panel attrition bias, although Pasek et al (2009) used standard multiple imputation via Amelia II (King et al, 2001). Deng et al (2013) outline the limitations of such approaches-both assume that the attrition is MAR.…”