Research Tips

Cluster Analysis and Latent Class Models

Researchers frequently want to create groupings based on survey responses or other types of data. For example, market segmentation involves grouping customers and potential customers based on product use, attitudes, and / or demographic characteristics. Similarly, a common component of competitive analysis involves grouping products, brands, and companies with those to which they are most similar. While groupings like these can be created by observing particular characteristics, cluster analysis and latent class models offer more sophisticated alternatives. There are many variations of cluster analysis and latent class models, but all involve using multivariate statistics to group things together based on simultaneously considering their similarity with respect to a large number of characteristics.