Direct or indirect? Assessing two approaches to the measurement of strategic voting