nettles9
I believe there’s a long list of Democrats because there are many that tossed their hat into the field. I don’t think there are many Republicans listed because there are less of them with hats in the field. If anyone has a list of Republicans that are running for president please share them. Thank you.
9:16 AM May 7th

TudorFever
I suspect that I am not alone in that I lean Republican and checked out of this process very early because it seemed to be just a laundry list of Democrats. Therefore I believe that this poll may be more useful in evaluating Democrats relative to each other than in evaluating Democrats versus Trump.​
10:58 PM May 6th

agcohen
How did you select this list of candidates? I don't think all of them are currently running for President, and there are several I have never heard of, and I follow politics to some extent. Some of the match-ups you list (and I only see them on the Bill James Online website, I don't use Twitter) seem ridiculous in the "candidates" they include.
5:48 PM May 6th

chrisbodig
This is fascinating. The fact that Kasich is doing well (with Weld and Schultz doing OK) tells me that, among your Twitter followers, you have quite a lot of independents or "Never Trump" Republicans.

I went through all of your polls and noticed that Trump, in 3 polls, got 27%, 28% and 26%. That's probably a fairly accurate reflection of the percentage of Trump supporters who follow you on Twitter.

One methodology question: does the total number of respondents in each poll factor into the equation somehow (I didn't notice a mention of it in the other article)?

It seems like most of the polls get between 1,000-1,600 voters but there are notable exceptions.

Your first poll (all heavyweights with Trump, Biden, Sanders and Harris) got a total of 2,497 votes, by far the most in your polls. Maybe that was just an initial "wow, Bill James is polling presidential candidates" factor or maybe it was because you had four high profile names that inspired a click.

When you had a poll of all lightweights (Weld, Messam, Gabbard, Williamson), only 615 people voted, making it a poll slightly less popular than the one in which you asked if people could tell the difference between a dove and a pigeon. Weld was the runaway winner (56%) in this poll. That tells me that a lot of Democrats didn't see any names they recognized (and didn't vote) but the "Never Trump" Republicans saw one of their people and clicked.

Anyway, I'm guessing the the total respondents per poll might render itself irrelevant by the time you get to poll #100 or #200.

Thanks for doing this, Bill. Will be a daily visit along with 538 and Real Clear Politics.


12:18 PM May 6th

myersb
When someone says 'I have absolutely no doubt that I am right', I tune out... nothing political in that: it's just obnoxious.
11:52 AM May 6th

taosjohn
Could you not get a pretty fair read on the validity of each sample by also asking them which of the four leading candidates they voted for last time? If the result of that poll varies very sharply from the actual outcome, that would suggest that your sample isn't all that representative to start with.

Not that it matters much because your system will rend to sort wheat from chaff anyway...


9:43 AM May 6th

phorton01
I guess I don't fully understand (or agree with) the methodology, but it seems to me it ignores the very inconvenient, but real, problem that the election will end being between one Republican and one Democrat. Giving people a choice between a mix of the parties seems like it will give skewed results.

I suppose you aren't necessarily trying to offer a prediction of what is likely to happen. Allowing Republicans to vote for their favorite Democrat (and vice versa) will not lead to insight into who will actually win the nomination.
9:05 AM May 6th