This paper empirically evaluates the usefulness of the Two-Step approach for transforming continuous variables toward normality. The study uses 27 corporate financial performance (CFP) variables on 39,216 US corporations to compare three variable sets: 1) random-normal, 2) original, and 3) transformed toward normality using the Two-Step. The results of several statistical procedures relevant to formative index (construct) construction are used to compare the three variable forms. The results provide strong evidence that the Two-Step approach is useful for 1) achieving normality improvements in continuous variables, 2) improve sampling adequacy for factor analysis, 3) dramatically increase intercorrelations, and 4) dramatically increase main effects tests involving the CFP variables. The findings have tremendous implications for MIS research and practice, as the Two-Step technique is shown here to change effects tests significantly and consequently has profound implications for the advancement of the MIS discipline and practical applications (e.g., data mining).