LIBERAL DEMOCRACY - VALIDITY AND METHOD FACTORS IN CROSS-NATIONAL MEASURES

This paper examines the definition and measurement of liberal democracy. Specifically, my purposes are (1) to propose a working definition of liberal democracy; (2) to outline a theory of ''method factors'' in subjective measures of liberal democracy; (3) to provide the first estimates of the proportion of variance due to systematic error, validity, and random error in commonly used measures; (4) to replicate these results across several years; and (5) to estimate the degree of liberal democracy in more than 150 countries. All but one measure contain systematic error, and in some cases the bias component is large. Furthermore, a new liberal democracy index has a .96 squared correlation with the liberal democracy latent variable and has negligible correlations with the method factors that are present in the individual indicators. The results suggest that the current practice of treating unadjusted democracy indicators as error free can be misleading.