= Parallel analysis
An approach to selecting factor or component count by comparing observed <eigenvalues> with reference values obtained from uncorrelated random data of matching dimensions. The reference must match the intended extraction and correlation approach. <John L. Horn> proposed a random-variable procedure to assess sampling contributions to the roots.<A rationale and test for the number of factors in factor analysis by John L. Horn>[1965, abstract]{ref} Use it with substantive interpretation and model assessment rather than as a universal guarantee of a correct factor count.
Back to article page
