Partitioned Variates: An Alternative to Factor Analysis.

摘要

Factor analysis has been widely used to understand interrelationships among sets of psychological variables, but this method might not be most appropriate for certain types of data analyses. Another approach using multiple correlation techniques is recommended as an alternative method which has more to offer the investigator under certain conditions. The method, originally, suggested by Guttman and here called partitioning, is contrasted with factor analysis and described in detail with examples. Partitioning takes into account the unique variance present in a variable and seeks to utilize this unique variance as meaningful measurement. (Author)