Fig. 1

Test scores are used to estimate the most likely scores on latent variables (constructs). In this example, three tests are used to measure latent variable 1 (e.g., emotion perception) and three more test scores are used to measure latent variable 2 (e.g., analytical intelligence). On the basis of the correlations between the tests, the most likely loadings of the tests on the latent variables are calculated, together with the most likely correlation between the latent variables. The latent variables are traits that cannot be observed directly, because every test is a combination of variance due to the latent variable and methodological variance