Factor Analysis, Random Data, and Patterned Results Academic Article uri icon

abstract

  • Multivariate statistical techniques such as factor analysis are capable of producing patterned results with most, if not all, data matrices. This paper demonstrates that patterned results are obtainable when principal component analysis is applied to a random data set. It is suggested that Bartlett's test for the statistical significance of a correlation matrix be employed in deciding whether a factor analysis of the matrix is justified.

published proceedings

  • American Antiquity

author list (cited authors)

  • Vierra, R. K., & Carlson, D. L

citation count

  • 18

complete list of authors

  • Vierra, Robert K||Carlson, David L

publication date

  • April 1981