Measurement versus Structural Model
In Structural Equation Modeling (SEM), we can think of a model in two perspectives:
- Measurement Model
- Structural Model
The measurement model deals with the relationship between a latent variable and its indicators.
For example: In measuring the ATTITUDE(ATT) of employees towards the ADOPTION INTENTION of new technology at the workplace, the researcher has four (4) and three (3) items with good loadings ( > 0.70) measuring respectively the latent variables ATTITUDE i.e. ATT1 - ATT4; and ADOPTION INTENTION i.e. AI1-AI3. So, the link between each of these items/indicators with their respective latent variable is known as a measurement model.
e1- e7 are the measurement error terms .
In contrast, a structural model defines the relationship between the various constructs in a model. In the example above, the two measurement model becomes a structural model when they are linked together as shown below. Thus, specifying how latent variables directly or indirectly affect other latent variables in the model.
res1 is the residual error term.