The effect of sampling error on convergence, improper solutions, and goodness-of-fit indices for maximum likelihood confirmatory factor analysis

Abstract

Confirmatory factory analysis, LISREL, Monte Carlo, Maximum likelihood,

Impact of noise on self-rated job satisfaction and health in open-plan offices: a structural equation modelling approach

“…Therefore, the proposed models were re-tested using separated data for China and Korea (N=192 for China and N=122 for Korea). It was assumed that the sample sizes for China and Korea were adequate, because previous studies had reported that a sample size of 100 is usually sufficient for convergence (Anderson and Gerbing, 1984;Iacobucci, 2010). Fit indices for both path models for the two countries showed acceptable levels (RMSEA = 0.06, GFI =0.90, and 2 /df=1.14 for China and RMSEA = 0.03, GFI =0.91, and 2 /df=1.89 for Korea).…”

Marketing research using single-item indicators in structural equation models

“…SEM is also important for validity testing and construct validation, because it focuses on the distinction between the measurement model and the structural model, but also allows more rigorous tests of construct reliability, convergent validity and discriminant validity (Anderson and Gerbing, 1984;Malhotra et al, 1999;Jarvis et al, 2003;MacKenzie et al, 2005).…”

Focus on opportunities as a mediator of the relationship between business owners' age and venture growth

“…We note that we reduced the number of parameters to be estimated by calculating a path analytic model and not a full structural equation model with a measurement and a structural model. We further used maximum-likelihood estimation which should also contribute to the robustness of our findings; Monte Carlo studies showed that bias in parameter estimates is of no practical importance for sample sizes as low as 50 in the case of maximum-likelihood estimations (Anderson & Gerbing, 1984;Gerbing & Anderson, 1985). Statistical inferences based on tests of significance remain valid because standard errors of the path coefficients are adjusted according to the sample size.…”