Here’s those graphs again, the way we’d be used to seeing them - with the supposedly dependent variable on the y-axis.
Looks sort of like blobs of nothing, except at the horizontal axis where something changes. It looks a lot more like, “Hardly matters how corrupt you are, but it sure looks like something weird is happening in part of the graph.”
There are a few effects here that I’m curious about. Maybe these were well controlled for, maybe not.
First, length of term. Your chances of reelection go down with every term, your level of corruption likely goes up with every term. So I wonder how much the “too corrupt” overlaps with “in power too long.”
Second, I wonder about those people who were not corrupt and who lost votes. I don’t know how the corruption index was calculated, but it can only be calculated based on information that is actually available. There tends to be more information available about people who people are paying attention to. Are we seeing an effect where people who don’t get much attention don’t get many votes and don’t register as corrupt because of the lack of attention?
