They show that the prespow1 confidence intervals for Latin America are generally smaller than for Africa. This means that scholars are in more agreement about presidential power scores for Latin America than for Africa. By contrast, we can see that the confidence intervals for Latin America tend to be larger for prespow2 than for prespow1, whereas they are smaller for Africa for prespow2 than for prespow1. So, while we do not make any specific recommendation, this suggests that a purely Latin American study might use the prespow1 scores because they are more reliable than the prespow2 scores, whereas a purely African study might use the prespow2 scores. A scholar wishing to do a large-n comparative study that included both regions would have to look at what particular countries were being included in the sample and whether prespow1 or prespow2 scores were on the whole more reliable for that.