When can categorical variables be treated as continuous? A comparison of robust continuous and categorical SEM estimation methods under suboptimal conditions.

By Mijke Rhemtulla, Patricia E. Brosseau‐Liard, Victoria Savalei. Psychological methods, 2012

2,457 citations recorded by the source corpus.

Subfields: Management Science and Operations Research, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/383333f4b5abdaf4f67e09343548f7e0930524cb

How researchers see this paper

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

In Management Science and Operations Research

In Statistics and Probability

Citations in the corpus

“My Questionnaire is Too Long!” The assessments of motivational-affective constructs with three-item and single-item measures (2014, methodology): “In other circumstances, more sophisticated methods that best fit the distributional properties of the measures under inves tigation should be chosen (see, e.g., the methods discussed by Rhemtulla et al. (2012) or Wirth and Edwards (2007)).”
Are Robust Standard Errors the Best Approach for Interval Estimation With Nonnormal Data in Structural Equation Modeling? (2018, methodology): “The Copula condition was exactly the same as the Severe condition in Study 1, whereas VM and CN were at levels intermediate to conditions in Study 1, In addition, two Likert-type data conditions were implemented in which multivariate normal data were discretized following Rhemtulla et al. (2012).”
Performance of Estimators for Confirmatory Factor Analysis of Ordinal Variables with Missing Data (2020, background): “In contrast, bias for MLR estimates did not appear to be affected by missingness but by skewness and discreteness of indicators as expected (e.g., Li, 2016; Rhemtulla et al., 2012).”
How True Is Grit? Assessing Its Relations to High School and College Students’ Personality Characteristics, Self-Regulation, Engagement, and Achievement (2017, background): “In a simulation study, Rhemtulla et al. (2012) found that treating categorical variables as continuous variables yielded unbiased or acceptable levels of parameter estimates and standard errors only when the number of categories was at least five and thresholds of item responses were symmetric.”
Associations Between Grit, Motivation, and Achievement in High School Students (2017, background): “In a simulation study, Rhemtulla et al. (2012) found that data should be treated as continuous only when there are at least five response options and the shape of the response distribution is symmetric.”
Dimensional comparisons: How academic track students’ achievements are related to their expectancy and value beliefs across multiple domains (2018, background): “structural parameter estimates (which were the major focus of our study) when there are at least four response categories and only moderate levels of nonnormality (Beauducel & Herzberg, 2006; DiStefano, 2002; Muthén & Kaplan, 1985; Rhemtulla et al., 2012).”
Testing Measurement Invariance in Longitudinal Data with Ordered-Categorical Measures (2016, positive): “…(e.g., from “I somewhat believe this” to “I very much believe this”) are not distributed symmetrically around the mean of the latent response distribution, such that the peak of the distribution of the observed ordinal indicator falls to the left or right of the center (Rhemtulla et al., 2012).”
Your Coefficient Alpha Is Probably Wrong, but Which Coefficient Omega Is Right? A Tutorial on Using R to Obtain Better Reliability Estimates (2020, methodology): “…data are to be handled with direct ML estimation (see Brown, 2015), and the MLR estimator requests ML with robust model-fit statistics (see Savalei, 2018), as recommended by Rhemtulla, Brosseau-Liard, and Savalei (2012) to account for nonnormality inherent in the analysis of item-level data.”