I’ve written before about the problems with Prof. Scheppele’s writing. It’s not scholarship that she produces. My post on “The Hungary ‘Expert’ Discredited on Twitter” shows her close relationship to senior-level operatives in Hungary’s opposition and how, neglecting any semblance of academic objectivity, she takes her information seemingly exclusively from sources closely affiliated with the political opposition.

And in “The Princeton Professor Fails Electoral Math,” I show how the professor doesn’t seem to understand the mathematics of a mixed electoral system, one that combines majoritarian rules and proportional rules.

There’s another major blunder in the latest post. Take a look at this. Trying once again to prove that the districts were gerrymandered, she writes:

"The relationship between district size (here measured by the number of voters who actually voted in each district) and partisanship (measured by the percentage of the vote that went to the left alliance and LMP combined) is substantial. As one would expect in a system gerrymandered to benefit the right wing of the political spectrum, the more left-leaning a district is, the more voters it has. In fact, the explained variation (R-squared) is 37%. In social science, it’s a huge effect when one variable accounts for more than one-third of the variation in another."

Did you catch that? Read again the first phrase in parentheses.

Why would anyone measure the size of an electoral district based on the number of people who turned out to vote on April 6th? The size of an electoral district – in Hungary, in Germany, in the United Kingdom, in the United States, or anywhere – is the number of eligible voters.

What she’s trying to show is that the relatively bigger districts are allegedly the districts that lean toward the opposition. She claims that the districts were deliberately drawn that way to give less weight to those opposition votes.

Sounds good, but why doesn’t she calculate the size of the district using the correct number – i.e., the total number of eligible voters? Well, because she can’t prove anything that way.

She claims that the explained variation (R-squared) is 37%. But what R-squared is really showing in her example is a correlation between voter turnout and success of the opposition parties. That is, in districts where there were more votes cast – not in districts that were bigger – the left fared better. But that’s no surprise. Even the opposition leader, Ferenc Gyurcsány, said prior to the elections that higher turnout figures would serve their interests.

So I looked at the numbers myself. What’s the value of R-squared if we use the correct number – i.e., number of eligible voters – to indicate the size of the district? Answer: 2.3%

That’s kind of a big mistake.

"In social science,” writes Scheppele, “it’s a huge effect when one variable accounts for more than one-third of the variation in another." Not only has she got the variable wrong, but she confuses correlation with causation. Any beginning student of statistics and social science knows that correlation does not imply causation. Maybe the professor has other intentions than using real scholarship to prove a point.

Does the Princeton professor make these mistakes and manipulate these numbers deliberately or are these simply blunders based on her lack of understanding? I’m not sure which is worse.

By Ferenc Kumin

Source: A Blog About Hungary

This opinion does not necessarily represent the views of this portal. Your opinion articles are welcome too, for review before possible publication, via info@xpatloop.com