Discussion about this post

User's avatar
Theodore Whitfield's avatar

Hey! I love me some ZH. But I think if you want to venture there then you have to be prepared to filter out the nutty stuff. Of course, that's true of everything.

I agree that it is bad practice to not start graphs at 0, and I agree that the ZH graph did indeed distort the magnitude of the effect. I also applaud you for your firm commitment to statistical communication best practices. However, to be fair a LOT of people commit this sin, and some will even be willing to explicitly justify the practice, precisely because it exaggerates the effect. In that case, it's not fair to pick on poor little ZH.

Even though the ZH graph distorts the actual effect, it's nonetheless true that there really was a remarkable increase in the number of voters in 2020, and it's not statistically consistent with the variation observed between the turnouts for the other elections.

Expand full comment
Dr. Ken Springer's avatar

Amazed that ZH would be so obviously deceptive. Tinkering with y-axis scaling is one of the oldest and widely used tricks. Darrell Huff complained about it in 1954 ("How to Lie with Statistics"), and you can still see it everywhere (Fox seems especially enamored).

Expand full comment
10 more comments...

No posts