Principal Component Analysis of Residuals (PCAR)

An added technique that can be used to investigate the dimensionality of a test or survey (e.g., do all items define the same trait?) is Principal Component Analysis of Residuals. In this chapter, we talk readers through the steps required to conduct a Principal Component Analysis of Residuals- from the eigenvalues that may need to be observed to support potential multidimensionality, to the steps that must be taken to verify that items flagged as possibly “off dimension” are indeed not a part of a single trait. This chapter links to the Fit chapter of Rasch Analysis in the Human Sciences, and to the chapter of Advances in Rasch Analysis in the Human Sciences concerning the important use of Cross plots for a Rasch analysis.