University of Southampton OCS (beta), CAA 2012

Font Size: 
Bootstrapped Correspondence Analysis in Practice
Kris Lockyear

Last modified: 2011-12-23


The concept of using bootstrapping to assess the stability of maps derived from Correspondence Analysis was first presented to CAA 25 years ago.  The technique has not, however, been widely adopted although it is discussed by Michael Baxter in his volumes on statistics in archaeology.  One probable reason is the lack, until recently, of easily obtained software to undertake such analyses.  Routines are now available to perform these analyses in the open-access statistical system R.  This paper applies bootstrapped CA to five archaeological data sets, and assesses how the technique enhances the interpretation of the results of CA.  The impact of sample size is discussed, and a comparison between methods is presented.


bootstrapping, Correspondence Analysis, multivariate statistics, data display, R