Using an EM Covariance Matrix to Estimate Structural Equation Models With Missing Data: Choosing an Adjusted Sample Size to Improve the Accuracy of Inferences

By Craig K. Enders, J. Peugh. 2004

131 citations recorded by the source corpus.

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

View Paper: https://www.semanticscholar.org/paper/679cb9f4fa27c11c57570c8fb11d664f5547ca53

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

Citations in the corpus

Robust Two-Stage Approach Outperforms Robust Full Information Maximum Likelihood With Incomplete Nonnormal Data (2014, methodology): “Some authors refer to these as the EM means and the EM covariance matrix (Enders & Peugh, 2004; Graham, 2003) because the likelihood in Equation 1 can be easily maximized by the EM algorithm.”
Spirituality in the Workplace: Theory and Measurement (2011, methodology): “Assuming data were missing at random, we used ExpectationMaximization algorithm to impute missing values (Allison, 2003; Enders, 2001, 2003, 2004; Enders & Peugh, 2004).”
The Influence of Childhood (2012, methodology): “Consequently, an expectationmaximization (EM) covariance matrix was estimated and maximum likelihood (ML) used to estimate model parameters and standard errors (Enders & Peugh, 2004).”
Exploring the journey of school improvement: classifying and analyzing patterns of change in school improvement processes and learning outcomes (2011, background): “One advantage of the latent curve analysis (LCA) approach is that missing data and student mobility can be incorporated directly into the analysis, which reduces parameter bias that could result from eliminating these cases (Enders & Peugh, 2004).”
Comparison of Methods of Handling Missing Time-Invariant Covariates in Latent Growth Models Under the Assumption of Missing Completely at Random (2007, methodology): “The second approach is to adjust the size of the sample used in the SEM analysis (e.g., Enders & Peugh, 2004).”
A Two-Stage Approach to Missing Data: Theory and Application to Auxiliary Variables (2009, methodology): “ Enders and Peugh (2004) concluded that, “Unfortunately, there is no single value of N that is appropriate when using an EM covariance matrix as input into an SEM analysis” (p. 18).”
A Statistically Justified Pairwise ML Method for Incomplete Nonnormal Data: A Comparison With Direct ML and Pairwise ADF (2005, methodology): “The term EM has also been used (e.g., Enders & Peugh, 2004) to describe a method that arrives at estimates of µ and Σ in the absence of any structural model and then estimates the model using these as if they were complete data statistics.”
On Obtaining Estimates of the Fraction of Missing Information From Full Information Maximum Likelihood (2012, background): “If the entire model is saturated (e.g., if the model is a regression model), O and O† are equal to the “saturated ML” mean and covariance matrix estimates typically printed in the output, also known as the EM means and covariances (Enders & Peugh, 2004; Graham, 2003).”
When can categorical variables be treated as continuous? A comparison of robust continuous and categorical SEM estimation methods under suboptimal conditions. (2012, background): “As such, 95% is the optimal value of coverage, and coverage below 90% is considered inadequate (Collins, Schafer, & Kam, 2001; Enders & Peugh, 2004).”