Estimating Measurement Error in Longitudinal Data Using the Longitudinal MultiTrait MultiError Approach

Abstract

Longitudinal data makes it possible to investigate change in time and its causes. While this type of data is getting more popular there is limited knowledge regarding the measurement errors involved, their stability in time and how they bias estimates of change. In this paper we apply a new method to estimate multiple types of errors concurrently, called the MultiTrait MultiError approach, to longitudinal data. This method uses a combination of experimental design and latent variable modelling to disentangle random error, social desirability, acquiescence and method effect. Using data collection from the Understanding Society Innovation Panel in the UK we investigate the stability of these measurement errors in three waves. Results show that while social desirability exhibits very high stability this is very low for method effects. Implications for social research is discussed.