Forfatter av avsnitt: Danielle J. Navarro and David R. Foxcroft

Faktoranalyse

Previous chapters have covered statistical tests for differences between two or more groups. However, sometimes when conducting research, we may wish to examine how multiple variables co-vary. That is, how they are related to each other and whether the patterns of relatedness suggest anything interesting and meaningful. For example, we are often interested in exploring whether there are any underlying unobserved latent factors that are represented by the observed, directly measured, variables in our data set. In statistics, latent factors are initially hidden variables that are not directly observed but are rather inferred (through statistical analysis) from other variables that are observed (directly measured). For a more technical and useful geometric explanation of the essentials of Factor Analysis take a look at Child (2006). Kline (1994) also provides a helpful introduction to the basic theory of Factor Analysis.

In this chapter we will cover how to undertake a number of different Factor Analysis and related techniques, starting with Eksplorerende faktoranalyse (EFA). EFA is a statistical technique for identifying underlying latent factors in a data set. In the next section, we will cover Prinsipal komponentanalyse (PCA) which is a data reduction technique which, strictly speaking, does not identify underlying latent factors. Instead, PCA simply produces a linear combination of observed variables. Following this, the section Bekreftende faktoranalyse (CFA) shows that, unlike EFA, with CFA you start with an idea – a model – of how the variables in your data are related to each other. You then test your model against the observed data and assess how good a fit the model is. Finally, we introduce Reliabilitetsanalyse av intern konsistens which tests how consistently a scale measures a psychological construct.