Structural equation modelling

The last section gives you the chance to try something a little different. We will be using a technique called structural equation modelling (SEM). SEM is a powerful statistical technique with a wide range of purposes, which allows one to simultaneously test several relationships at once, including factor structures, regression coefficients and between group differences. It also allows the incorporation of latent variables into models. SEM encourages confirmatory rather than exploratory approaches, and so is seen as a deductive scientific approach. Rather than having a set of data and seeing what is thrown up, using SEM effectively requires one to have a testable model before coming to the data. This makes it different from standard exploratory factor analysis, and some approaches to regression.

We will be using a package called AMOS, which is part of the SPSS suite of software. Not all of you will have access to this package.

The specific question we will be asking comes from a paper published on the relationship between well-being, income and aspirations1 (more on aspirations). Specifically, the paper argues that aspiratons mediate the relationship between income and well-being, such that for people who have more extrinsic aspirations, income will be more important in determining well-being, whilst for people who have more intrinsic aspirations, well-being will not be so determined by income. The paper found, using regression techniques, that there was indeed an interaction between income and aspirations in the determination of well-being, as operationalised by life satisfaction.

We will attempt to confirm these findings using a more robust methodology, which is possible through SEM (Multiple Group Analysis). Aspirations might have the opposite effect on a different independent variable. Perhaps, if income is less important for less materialistic individuals, then other determinants might be more important instead.

Go to next page >>

Footnotes