A factor analysis is a method of multivariate statistics in which numerous variables are reduced to only a few relevant factors. It belongs to the structure-discovering methods and serves to reduce complexity. As a rule, the so-called manifest variables are based on collected data, while the latent factors are not collected constructs. The reduction of the number of variables can take place either through confirmatory or exploratory factor analysis:

For example, factor analysis can be used to assign different image items to superordinate dimensions (such as assessed friendliness or competence of employees to the dimension of customer service) or different products to product families that are consistent in terms of content. Furthermore, already existing allocations can be checked by means of factor analysis.

Factor analysis also serves to validate latent constructs within other multivariate procedures (see for example structural equation model) or serves to identify so-called supervariables, which are relevant to explanation for segmentation, for example.