Factor Analysis

Introduction

In situations where many variables are at play, factor analysis identifies the key underlying factors or constructs within a set of data. A factor analysis may start with twenty known (or manifest) variables from a questionnaire and reduce them to, for instance, five constructs, or latent factors.

A greater understanding of these latent variables allows one to form a more nuanced understanding of a relationship than linear or multiple regression analysis. Factor analysis aids in understanding complex concepts that are difficult to measure directly, like socioeconomic status and personal beliefs.

The main steps of factor analysis include:

  1. Determining your variables. In a marketing context, these generally are product descriptions from the customer’s perspective.
  2. Collecting preferences via questionnaires or surveys.
  3. Performing factor analysis (in most instances, this involves running the data through a computer-based model) to determine underlying factors.
  4. Using the underlying factors to inform brand and product decisions.

Factor analysis tends to be most effective with larger sample sizes: 300 is good, 500 is very good, and 1000 or more is excellent (Comrey and Lee 1992).

Examples

A CPG company wants to segment the consumer base for their baby food offerings. Below are some hypotheses about needs-related attitudes of parents choosing food for their baby. These variables can be condensed into the following underlying factors:

Factor Analysis

Applications

Factor analysis in psychology is mainly used for intelligence research, but has also been used to find factors in a broader range of areas such as personality, attitudes, beliefs, etc. Within a marketing context, factor analysis can be particularly useful in constructing perceptual maps and brand positioning.