A primer for the estimation of structural equation models in the presence of missing data: Maximum likelihood algorithms

Maximum likelihood algorithms have undergone substantial development, yet marketers have been slow to adopt these techniques to address missing data. This study familiarises marketers with the available maximum likelihood estimators, reviews missing data theory and research, and presents a structural equation modelling simulation study to demonstrate the advantages of maximum likelihood estimation versus other techniques. Results indicate that the full information maximum likelihood and expectation-maximisation outperform traditional techniques with respect to parameter estimate bias, model fit and parameter estimate efficiency. Marketers should be aware of the potential impact of missing data assumptions and decrease their reliance on ad hoc methods in favour of maximum likelihood estimators.