https://doi.org/10.21697/sp.2017.17.1.04
The aim of this article is to discuss the reconstruction of the theoretical elements of the structures verified by structural equation models (SEM). The elements of the structures have single-level (one-dimensional) or hierarchical (multi-dimensional) characteristic. The reconstruction of the characteristics of the theoretical elements is very important issue. Poor reconstruction of the theoretical elements, as will be shown, may increase the likelihood of the error of the first kind and lead to the rejection of the correct model. There are such elements which are so complex that verification of the models with their participation would require a huge effort involving trial and finance. If the research concentrates merely on the reconstruction of the characteristics of a particular element, then its construction is crucial. But when the research concentrates on the verification of the model of the process, in which the elements are plentiful and they are connected in a network of mutual dependencies, complexity of the model grows to such an extent that may require thousandths of observations and this can lead to a problem in its estimation. In this case, it becomes necessary to reduce the characteristics of the element to a meta-feature. The article explains how to verify the correctness of the construction of a one-level latent variable, hierarchical latent variable and estimate the variance extracted and reliability of variables. There is also shown how to reduce the hierarchical variable to meta-feature in order to simplify the model and reduce the number of degrees of freedom. There is presented how the construction of meta-features translates into both the reliability of the variable and the fit of the model. The construction of the variables is discussed on the example of the scale of Reaction to stress and the scale of Constraining child’s activity.
Download files
Citation rules
Cited by / Share
English