It is argued that construct validity is content validity and that prior content validity is necessary for predictive validity. Too many measures of abstract (multicomponential) constructs in management and organizational research exhibit poor content validity or else lose much of their content validity following unnecessary statistical ‘purification’ to select items. The measures produce misleading empirical results and can lead to erroneous acceptance or rejection of hypotheses or entire theories. Both problems – inadequate content validation and unnecessary statistical purification – are illustrated here for two new measures of the construct of ‘export coordination’ (Diamantopoulos and Siguaw, British Journal of Management, 2006, 17 (4), pp. 263–282). Corrections according to Rossiter’s C-OAR-SE procedure for scale development are specified.