In addition, group differences can be modelled, through differences in means, variances and variance components, and through differences in the way symptoms relate to the latent trait. For instance, one or more symptoms may show a higher incidence rate in one group (indicated by a difference in β-parameters across groups, e.g., females and males), or be a more sensitive indicator for the trait in a particular group (indicated by a difference in α-parameters across groups). Such violations of measurement invariance are usually referred to as differential item functioning (DIF).
