Fixed loss of significance error in PersonsCorrelation p-value computation causing p-values smaller than the machine epsilon (~1E-16) to be reported as 0. JIRA: MATH-371 Reported and patched by Kevin Childs