Xinya Liang

An evaluation of WLSMV and Bayesian methods for confirmatory factor analysis with categorical indicators

This Monte Carlo study evaluated the performance of three estimation methods in fitting confirmatory factor analysis (CFA) models for ordered categorical data, with the focus on data with non-normal underlying distributions and small sample sizes. The three methods are: weighted least squares with mean and variance adjusted (WLSMV), Bayesian with non-informative priors (BN), and Bayesian with informative priors (BI). Design factors included sample sizes, factor structures, underlying continuous distributions, and categorical distributions. Results were evaluated based on the model-data fit, point estimates, and standard errors of point estimates. Results showed that Bayesian methods encountered less convergence problems than WLSMV. Bayesian methods were robust to the non-normality of underlying continuous distributions. WLSMV tended to perform equally well or slightly better than Bayesian methods except for some conditions with small sample sizes or highly non-normal underlying distributions.

Effect of worked examples on mental model progression in a computer-based simulation learning environment

In a diagnostic problem solving operation of a computer-simulated chemical plant, chemical engineering students were randomly assigned to two groups: one studying product-oriented worked examples, the other practicing conventional problem solving. Effects of these instructional strategies on the progression of learners’ mental models were examined by comparing representations of their mental models with those of experts at three segments of the instruction. Progression of mental models for the worked example group was significantly greater than those using the problem-solving strategy. However, this progression did not manifest itself in learners’ troubleshooting performance measured by number of correct diagnosis and first time correct diagnosis. The implications of these results for designing instruction tailored to learners’ domain knowledge are discussed.

Confirmatory factor analysis under violations of distributional and structural assumptions

This simulation study evaluated CFA model results under violations of both distributional and structural assumptions using maximum likelihood (ML), robust maximum likelihood (RML), and weighted least square (WLS) estimation methods. Design factors included model complexity, the degree of non-normality of factor and error scores, sample sizes, and model misspecifications. In total, 72 conditions were used for data generation. Results were evaluated by rejection rate based on the model chi-square tests, fit function, CFI, RMSEA, and parameter estimates. Findings from the simulation study suggested that CFA results were robust to moderate violation of the non-normality of both factor and error scores under ML and RML methods, however, the degree of non-normality of factor scores impacted both overall model fit indices and loading estimates under WLS, particularly when the models were mis-specified. In addition, correctly specified and mis-specified models were likely detected by combining results from multiple estimation methods.

Evaluation of Structural Relationships in Autoregressive Cross-Lagged Models Under Longitudinal Approximate Invariance:A Bayesian Analysis

To infer longitudinal relationships among latent factors, traditional analyses assume that the measurement model is invariant across measurement occasions. Alternative to placing cross-occasion equality constraints on parameters, approximate measurement invariance (MI) can be analyzed by specifying informative priors on parameter differences between occasions. This study evaluated the estimation of structural coefficients in multiple-indicator autoregressive cross-lagged models under various conditions of approximate MI using Bayesian structural equation modeling. Design factors included factor structures, conditions of non-invariance, sizes of structural coefficients, and sample sizes. Models were analyzed using two sets of small-variance priors on select model parameters. Results showed that autoregressive coefficient estimates were more accurate for the mixed pattern than the decreasing pattern of non-invariance. When a model included cross-loadings, an interaction was found between the cross-lagged estimates and the non-invariance conditions. Implications of findings and future research directions are discussed.

Challenges and Opportunities for International Students in Graduate Education

International students pursuing graduate education in U.S. institutes have been rapidly increasing in recent years. Students from all over the world remarkably contribute to the advancement of U.S. economy and technology. This article addresses the challenges and opportunities international students face during and after graduate education. The challenges include overcoming the barriers with regard to language, culture, and employment. Along with these challenges are numerous opportunities such as introducing a different culture to U.S. college campuses, facilitating long-term academic and business collaboration, and obtaining enriched working experiences in the global job market. Strategies and recommendations to attain academic and occupational success are also discussed.