China 
Africa 
Explore chapters and articles related to this topic
Latent Variable Models
Published in Simon Washington, Matthew Karlaftis, Fred Mannering, Panagiotis Anastasopoulos, Statistical and Econometric Methods for Transportation Data Analysis, 2020
Simon Washington, Matthew Karlaftis, Fred Mannering, Panagiotis Anastasopoulos
Several important concepts are routinely applied throughout SEM GOF tests that enable the assessment of statistical models. A saturated model is a model that is perfectly fit to the data—the variance-covariance structure is completely unconstrained and represents an unappealing model. It is the most general model possible and is used as a standard of comparison to the estimated model. Because the saturated model is as complex as the original data, it does not summarize the data into succinct and useful relationships. In contrast, the independence model is constrained such that no relationships exist in the data and all variables in the model are independent of each other. This model presents the “worst case” model. The saturated and independence models are typically viewed as two extremes within which the best model lies.
Structural equation modelling of construction project performance based on coordination factors
Published in Cogent Engineering, 2020
Wesam Salah Alaloul, Mohd Shahir Liew, Noor Amila Wan Zawawi, Bashar S Mohammed, Musa Adamu, Muhammad Ali Musharat
The primary hypothesized model shown in Figure 5 was analyzed using AMOS 20.0.0 software. The purpose of adopting AMOS is due to its ability to provide results for saturated and independent models. The saturated model consists of all possible relationships which are defined, whereas the independent model contains estimated minimum numbers of parameters. Hence, this indicates that the two models stand at extreme points, while any proposed model can be in between these two models. In the case of the current research, GOF indicators of the last model managed to satisfy the suggested cutoff points after three iterations as well as deletion and addition of some covariances. Apart from that, various measures of goodness of fit exist; however, all SEM specialists tend to assess the models by calculating more than one of these measures. Kline (2015) recommended that the principles for best-fit measures are to be independent of sample size, accurate, and consistent in evaluating the dissimilarity models. Accordingly, the no-normed fit index (NNFI), comparative fit index (CFI), and the root mean squared approximation of error (RMSEA) were suggested.