Paul Vogt, author of the "Dictionary of Statistics and Methodology," defines principal component analysis as a data reduction technique that transforms a large set of correlated variables into a smaller group of uncorrelated variables. It is closely related to factor analysis and is often the first step in a factor analysis study. Principal component and factor analysis assume that one or more unmeasured factors influence a set of observed measures, such as a set of responses to survey questions.
Principal component analysis extracts the underlying factors in a set of data. The Statsoft website notes that analysts often determine the number of factors based on the eigenvalues, statistics that measure the amount of variation in the data. Many principal component analyses use a criterion in which the number of underlying factors equals the number of eigenvalues with values greater than 1.
Because of the complexity of the procedure, principal component factor analysis requires the use of a statistical software program, such as SAS or SPSS. Excel requires a statistical add-on program, such as XLSTAT, to conduct this type of analysis.