Trees show the relationship between two sets of questions as an  upside-down tree.

While most tree-based methods, such as CHAID and CART only predict a single categorical or numeric variable, Q’s approach can predict all the different types of data: numeric, categorical, rankings, multiple response, conjoint and other types of experiments.   

The figure below shows Q’s graphical representation of segmentation, with bars showing categorical data and histograms representing numeric data.  

In Q4, trees are created by clicking on the Create menu, selecting Segments, selecting Split by questions (tree) and choose the questions to split by.  Contact support@q-researchsoftware.com for a free upgrade to Q4.

An online tutorial explains how to create trees.  From the Help menu in Q, select Online Training, sign in, then go to Multivariate Techniques, Segments and Trees.

Latent class analysis finds groups in data.  It is most commonly used for market segmentation. 

Latent class analysis is a big improvement on cluster analysis.  It automatically deals with missing data; even if there are lots of “don’t know” responses, each person is still assigned to a segment.  It accommodates all the different types of data: numeric, categorical, rankings, multiple response, conjoint and other types of experiments.   

The figure below show’s Q’s graphical representation of segmentation, with bars showing categorical data and histograms representing numeric data. 

In Q4, latent class analysis is conducted by clicking on the Create menu and selecting Segments; contact support@q-researchsoftware.com for a free upgrade to Q4.

An online tutorial explains how to use latent class analysis.  From the Help menu in Q, select Online Training, sign in, then go Multivariate Techniques, Segments and Latent Class Analysis.