SEM includes a set of analyses used when one wants to see not only the relationship between measured (i.e. observed) variables and latent constructs (as in factor analysis), but also uncover the pattern, path, or underlying "structure" of a set of latent constructs. In simple terms, using observed variables to understand how unobserved constructs relate to one another.

Most of the times, theoretical models are written as SEM models and are inadequately analyzed with GLM (regression, ANOVA, loglinear models).

One mayor reason for using SEM models is when independent variables are correlated. These correlations are adequately dealt with in SEM models, but GLM assumed orthogonal variables.

Dear Anak Agung Bagus Wirayuda

Generally, speaking the structural equation modelling is a confirmatory approach, in which you have a theory of certain theory or a behaviour or an intention towards a particular behaviour as an intention to purchase, intention to purchase, intention to accept, and so on. The behaviour in question is so often influenced by a set of so-called factors or dimensions or sometimes construct, which are so often are latent variables. In other words, they are not measured directly, e.g. happiness, trust, satisfaction and others. This leads to the point of how the SEM is used.

SEM consists of two parts, which are the following:

Structural Model:

The researcher typically presents a picture of the expected relations among the selected factors, dimensions, or constructs interact with each other. These relations are usually based on the literature review, prior experience, or expected outcome of the individual's tendency.

Measurement Model:

The researcher asks a set of items or questions about each factor, dimension or a construct. They are used to capture the concept of the developed constructs. The aim source to acquire them is so-called dimension adaption, in which with an extensive review of recently published literature that matches the research context, the researcher should identify the items or questions that are likely to be selected. The adaption implies that some changes will go on the original scale.

Note:

However, I need to clearly mention that if you aim to develop a new scale, it is also possible. Still, it will take a longer time, and usually, it includes using the dimension reduction measures as Explanatory Factor Analysis (EFA) or Principal Component Analysis (PCA). Most of the studies do not develop a new scale for their construct; however, they rely on an already developed, reliable, and valid scale.

Finally, I have included three photos, which illustrate structural and measurement models. Also, I have put a diagram to walk you through the independent and dependent variables requirements to be sufficient to run SEM.

Best Regards,

Belal Edries

References:

Structural Model and Measurement Model

• Xu, Z., Zhang, K., Min, H., Wang, Z., Zhao, X., & Liu, P. (2018). What drives people to accept automated vehicles? Findings from a field experiment. Transportation Research Part C: Emerging Technologies, 95, 320–334. https://doi.org/10.1016/j.trc.2018.07.024

Research Method Selection

• Joseph F. Hair, William C. Black, Barry J. Babin, Rolph E. Anderson Multivariate Data Analysis 7th Edition 2009