On the interpretation of factor analysis.

The importance of the researcher's interpretation of factor analysis is illustrated by means of an example. The results from this example appear to be meaningful and easily interpreted. The example omits any measure of reliability or validity. If a measure of reliability had been included, it would have indicated the worthlessness of the results. A survey of 46 recent papers from 6 journals supported the claim that the example is typical, two-thirds of the papers provide no measure of reliability. In fact, some papers did not even provide sufficient information to allow for replication. To improve the current situation some measure of factor reliability should accompany applied studies that utilize factor analysis. Three operational approaches are suggested for obtaining measures of factor reliability: use of split samples, Monte Carlo simulation, and a priori models. With the advent of computers, factor analysis has become a relatively inexpensive technique. As a result, the number of published studies employing this method of analysis is rapidly increasing. Solomon (1960) reported that over 1,000 papers on factor analysis were published between 1900 and 1960 and that the rate of publication has increased steadily. These publications have been divided between papers on methods and papers on applied problems which utilize factor analysis as the primary statistical technique (Guilford, 1961). This paper is concerned with the latter category. Do the applied studies lead to advances in the description or understanding of real world situations?