Just Sharing

I would like to share a few thoughts about the “events” occurring around us personally and around the world. I do not have a problem with how other live their lives and what measures they take to…

Smartphone

独家优惠奖金 100% 高达 1 BTC + 180 免费旋转




A Tale of 3 Primaries

Who is winning the democratic primary? Depends who you ask.

There are three useful ways to look at a primary election. (That is, 3 ways better than asking talking heads.) Traditional polls (as represented by RealClearPolitics), forecast models (Nate Silver’s 538), and prediction markets (Predict It), usually the stories they tell are closely linked. However, lately they have diverged into distinct pictures. How the following weeks play out will shed light on who voters want, and who on understands that better. So what do they say?

Bernie: All have him leading the pack, but none describe it as his race to lose

Biden: Polls show him just barely overtaken by Bernie, 538 shows him in a distant 2nd, and PredictIt has him down in the 2nd tier alongside Mayor Pete and Amy Klobuchar

Bloomberg: Polls show him catching up to Biden, but only just overtaking Warren. He barely registers in 538, but PredictIt has him comfortably in 2nd. He is perhaps the biggest disagreement between the methods.

Warren: Well below her peak in the top-tier months ago, she polls in a 2nd-tier mix with Bloomberg and Buttigieg, but 538 doesn’t see her winning many delegates, and PredictIt has effectively written her off.

Buttigieg: The co-winner of Iowa and ‘strong 2nd’ finisher of NH is another point of disagreement in the methods. Polls show him rising, but still at the bottom of the 2nd tier. 538 and PredictIt put him at the top of the 2nd tier.

Klobuchar: The ‘strong 3rd’ finisher in NH surprised many and has seen her fortunes rise recently. Polls still put her at the bottom. 538 gives her no chance of winning the most delegates. But Predictit gives her a small but real chance of winning at the bottom of the 2nd tier.

No One: There is no good way to poll for a brokered convention, but if no one gets high enough in the polls, then they can’t win outright, and the nominee will be decided by delegates in July. Both PredictIt and 538 have this as the most likely outcome now. Some of the discrepancy between those two models could be because of how they account for what happens at a brokered convention. 538 has no way of assessing how likely a Bloomberg win at a convention would be, and PredictIt doesn’t have a market for “Bloomberg to win, but only at a brokered convention” so we are left to speculate.

Analysis

Polls: These are still the best method for getting at the ground truth of the present state. All the other methods rely primarily on these, and can only disagree with the OG method for so long before gravity pulls them back to earth.

Forecasters: Something is happening with the 538 model. It’s better than previous years that obsessed too much on some book Silver read (The Party Decides) overemphasizing the importance of endorsements, but it has Bloomberg way too low. The only explanation I can think of is that it’s too wedded to the most recent polling.

Prediction Markets: Theoretically, any pollster or forecaster could jump on a market and put their money where their model is. The market should average out all those models to the most accurate picture possible. In reality, these are subject to over-interpretation of noise and gambler’s fallacies. However, they are the only method that can give real-time feedback of say, a good debate performance, or a devastating new story, without waiting for new polling. This is also the best method incorporating expectations, like someone uniting the moderate wing of the party, and that somebody likely being Bloomberg.

Conclusion

As 538 and RCP figured out that aggregating polls is a better method than arguing about single polls, perhaps the best way to get a clear picture of the primary isn’t through one best method, but through looking a what each method has to say. Polls tell us where we are. Forecasters tell us where we are likely to go using the best math available. And Prediction Markets fill in the gaps that neither of the other methods can answer.

Add a comment

Related posts:

Privacy Protector

For all the newbies out there, Privacy Protector (PP) is an instructional series meant to inform the technologically challenged on the basics of keeping information as private as possible while…

How a Website Helps You in Your Business?

Web design refers to the design of the webpages we see on different websites. A website contains different types of pages. Web design is the sum of appearance and layout of the pages, and contents…

Flutter Community Rocks Droidcon London!

For several weeks we wondered if Flutter Community would be welcomed or shunned at Droidcon London this past Thursday and Friday. As fate would have it, we didn’t need to wait until the event to find…