There are now many studies that operationalize a presidential power variable. However, there are multiple and competing measures of presidential power. We identify 30 separate measures. These measures differ in the number and selection of countries that are scored and the time period covered. Moreover, given each measure is based on a different set of component powers, even when different measures score the same country, they often provide a substantively different presidential power score overall. To what extent are existing findings robust to the application of alternative measures? We show that there is great variation in the correlations between the country scores and demonstrate the empirical effect of such variation by replicating a study that employs a given measure of presidential power using alternative measures. The replication shows that findings can be highly sensitive to the particular measure of presidential power that is used. We aim to maximize the reliability of presidential power scores for a large number of countries over time. For each of the 30 measures we normalize the presidential power scores for each country. However, given the number of times a country has been scored across the different measures varies considerably, we impute the missing scores across 60 newly generated datasets and perform principal component analysis across all the measures. The resultant rotated factor scores serve as a comparable and cross-national measure of presidential power for country i at time t. This strategy generates a set of presidential power scores for a larger number of countries than currently exists for any single measure and indicates the general validity of any particular country’s presidential power score. We report the adjusted presidential power scores for all countries and discuss the extent to which established findings are likely to be called into question by the new dataset.
