Clustering can be thought of as the obverse of factoring. In factor analysis, observations are correlated with each other, and the correlation matrix is examined to see which items covary among themselves and which therefore can be combined into a simpler structure of factors composed of items. In cluster analysis, the correlation matrix is examined to see which individuals or observations covary among themselves and can be represented by clusters composed of different observations. Once the groups of individuals are composed, they are compared to each other in order to see if some external correlate exists.
For example, a large group of patients might be administered a series of cognitive measures. The scores on those measures are examined to see which individuals seem to be most similar to each other and can be clustered. Once the groups are empirically formed, their characteristics are examined to see if age or diagnosis or injury severity...
References and Readings
- Kaufman, L., & Rousseeuw, P. (2005). Finding groups in data: An introduction to cluster analysis. Hoboken: Wiley.Google Scholar