The comparative efficacy of imputation methods for missing data in structural equation modeling

By A. Olinsky, Shaw K. Chen, Lisa L. Harlow. European Journal of Operational Research, 2003

366 citations recorded by the source corpus.

Abstract Missing data is a problem that permeates much of the research being done today. Traditional techniques for replacing missing values may have serious limitations. Recent developments in computing allow more sophisticated techniques to be used. This paper compares the efficacy of five current, and promising, methods that can be used to deal with missing data. This efficacy will be judged by examining the percent of bias in estimating parameters. The focus of this paper is on structural equation modeling (SEM), a popular statistical technique, which subsumes many of the traditional statistical procedures. To make the comparison, this paper examines a full structural equation model that is generated by simulation in accord with previous research. The five techniques used for comparison are expectation maximization (EM), full information maximum likelihood (FIML), mean substitution (Mean), multiple imputation (MI), and regression imputation (Regression). All of these techniques, other than FIML, impute missing data and result in a complete dataset that can be used by researchers for other research. FIML, on the other hand, can still estimate the parameters of the model. The study involves two levels of sample size (100 and 500) and seven levels of incomplete data (2%, 4%, 8%, 12%, 16%, 24%, and 32% missing completely at random). After extensive bootstrapping and simulation, the results indicate that FIML is a superior method in the estimation of most different types of parameters in a SEM format. Furthermore, MI is found to be superior in the estimation of standard errors. Multiple imputation (MI) also is an excellent estimator, with the exception of datasets with over 24% missing information. Considering the fact that FIML is a direct method and does not actually impute the missing data, whereas MI does, and can yield a complete set of data for the researcher to analyze, we conclude that MI, because of its theoretical and distributional underpinnings, is probably most promising for future applications in this field.

Subfields: Computer Networks and Communications, Management Science and Operations Research, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/42ac31d8c4d207e29550bd98d787d565a7edbd83

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Networks and Communications

In Management Science and Operations Research

In Statistics and Probability

Citations in the corpus

Best practices for missing data management in counseling psychology. (2010, methodology): “Simulation studies comparing FIML with other imputation techniques demonstrate that the FIML procedure produces approximately unbiased results across a variety of parameter estimates, particularly at small sample sizes (N 100), and produces results similar to those for both EM and MI (Enders & Banda”
Motivating Employees of the Public Sector: Does Public Service Motivation Matter? (2010, methodology): “It avoids a reduction of the sample caused by the covariance matrix creation and the exclusion of cases, either with likewise or pairwise deletion (Wothke and Arbuckle 1996 ), and is superior to data imputation methods (Olinsky, Chen, and Harlow 2003 ).”
Tourism Entrepreneurship Performance (2015, methodology): “Thus, as recommended by Olinsky, Chen, and Harlow (2003), the nearest neighborhood approach was used to impute the missing entries prior to running exploratory and model fit analyses.”
Achievements and self-concepts in a comparison of math and science: exploring the internal/external frame of reference model across 28 countries (2008, methodology): “The problem of missing data was addressed by the use of multiple imputation (Olinsky, Chen, & Harlow, 2003 ; Schafer, 1997 ) using the LISREL software before the major analyses were conducted.”
Subjective norms, attitudes and intentions of Finnish consumers in buying organic food (2005, methodology): “To avoid the problem with missing values, full information maximum likelihood (FIML) method of estimation (see Olinsky et al., 2003) was used.”
Measuring Loneliness Among Middle-Aged and Older Adults: The UCLA and de Jong Gierveld Loneliness Scales (2014, methodology): “It is conceptually similar to regression imputation and produces results similar to multiple imputation methods (see Graham et al. 1996; Olinsky et al. 2003).”
Determinants and consequences of internal control in firms: a contingency theory based analysis (2010, methodology): “The EM iterates through a process of estimating missing data and then estimating parameters ( Olinsky et al. 2003 ).”
Re-examining the relationships among perceived quality, value, satisfaction, and destination loyalty (2018, methodology): “Missing values on variables from the data set were imputed through the nearest neighborhood approach (Olinsky et al., 2003).”