I am running an exploratory factor analysis on ordinal data, and used a polychoric correlation matrix with a small continuity correction of .1 to adjust for 0 values. This w