Errata

Hi everyone, I realized that I made a mistake in my recent post Scoring 2024 Election Forecasts, so I’m writing to correct it and apologize.

In the post, I tried to be a bit fancy and high-tech, so for the Brier score tables rather than just pasting in a PNG image, I created the tables with an online data visualization tool called Datawrapper and embedded them in the post.

This worked ok for the actual post on Substack, so if you actually click on the link and look at the post, the numbers in those tables are all correct.

But it came to my attention today (thank you to the vigilant reader who noticed this) that in the email version of the post that got sent out to everyone, the numbers in these tables are completely wrong. For example, here is what the first Brier score table looks like in the actual post on Substack (correct version):

But then here is that same table, in the email version (incorrect numbers):

I think what happened was that when I was drafting the post (before the election), I made these tables on Datawrapper with some hypothetical results just to get a draft and see what the table would look like. Then, once the results were in, I plugged in the real numbers and re-saved/re-published on Datawrapper. At that point things looked fine on my end, with the embedding in the Substack post showing the updated, correct numbers, but the version of the post in the email version that got sent out apparently showed the old version of the Datawrapper table rather than the current version.

Here’s some actual proof of this so you guys don’t think I’m just making up an excuse:

The current version of the embedded Datawrapper table at the time I sent out the post. The last edit was November 29, when I was pretty sure how that last House race would turn out (although I waited until it was officially called before clicking “send”).

An earlier draft version of the Datawrapper table, made before the election with incorrect numbers just to see what the table looked like (also without the heatmap colors). But this is the version that showed up in the email version of the post.

The other Brier score table also had incorrect numbers in the email version for the same reason. The contest winners table had the correct numbers in the email version, but the third-place winner was incorrectly referred to as “Anonymous Person” because the first saved draft of that Datawrapper table was before I had gotten official permission from that guy (Frederik Holzer) to publish his name.

Anyway I’m very sorry about that! In the future I’m going to avoid using these Datawrapper embeddings for tables, and just go back to the old-fashion method of typing the results into Excel and saving them as an image.

Thanks for reading!