Poll Results 5-10-2019

            Well, we can write off the theory that John Kasich has only done well in these polls because he was the only Republican on the list.   Somebody had suggested that for why Kasich was doing as well in my polls as he has, but yesterday I polled Kasich against Trump and two minor Democrats.   The predicted results for the poll, based on previous polls, would be that Kasich would beat Trump, 45-42.  In fact, he walloped Trump, 51-24.    That is much more under-performance by Trump than it is over-performance by Kasich, but what it suggests is that, in fact, TRUMP was only doing as well as he had in the previous polls because he was the only Republican on the list, but that Republicans, given a choice, would prefer Kasich.  

            Of course, in yesterday’s poll, the two Democrats also over-performed, and no doubt their "scores" from yesterday are misleading/inaccurate, and will be out of line with their real support.   We have Michael Bennet now pegged at 185; he’s probably not actually that strong, and I know that, but I also know that my system will adjust for that and correct his score with additional polling.   It’s like a basketball rankings.   Sometimes a team that isn’t that good will go 18-for-26 shooting three-pointers, just as a fluke, and beat a much better team.   It screws up the rankings for a while, but it happened.   Michael Bennet doing unrealistically well because he was the stronger Democrat in the field is like a basketball team overperforming in a single game because they have a 7-foot center, not really that good, but they happened to be matched up against a pretty good team whose regular center was injured/not able to play so they were using a 6-8 backup at center, so your 7-footer scored 30 points and you won 72-64.  Same thing; matchups matter in the short run.   But we’ll poll everybody repeatedly, and after 5 or 6 polls we’ll actually know where everybody is and how they are doing.

            Back to Kasich, we have now polled him five times.  The first time, April 25, he was matched up against Gillibrand, Cory Booker and Klobuchar.  He finished first in the poll with 37%, but his "score" from that poll, evaluated by all of the information that we know have, is only 558—his worst performance in any of the five polls.     We polled him the second time on April 30.   He was matched up in that poll against Biden, Delaney and Moulton, and he got only 33%, far behind Biden with 60%, but his "score" from that poll is 627.   We polled him the third time on May third, a week ago today, against Elizabeth Warren, Beto O’Rourke and Mike Gravel.    He got 32% in that poll, Warren got 47%, but that’s actually a really strong performance against Elizabeth Warren, who is first in my polling and will move further ahead tomorrow, since she is over-polling significantly in the poll that is currently running.    His score from that poll, in isolation, would be 718—the best that he has done. 

            We polled him again on Sunday, against Buttigieg, Klobuchar and Gillibrand.   He got 30% compared to Buttigieg’s 43%, but again, Buttigieg was much weaker in that poll than in most of the others.   Based on that poll, Kasich would get a score of 679.     Yesterday, thumping Trump 51-24, he gets a score of 676.   So he’s now been polled against Biden, Warren, Buttigieg. Beto and Trump, four or five heavyweight contenders, and he’s done reasonably well every time. 

            These are the updated standings:

            Marianne Williamson, who is dead last in my polling, has qualified for the Democratic debate stage based on the number of individual donors.   I don’t know much about her.   I gather that she is an author.  

            Anyway, Trump dropped 96 points and dropped from 6^th to 8^th in the rankings based on his very weak showing yesterday against Kasich and two minor Democrats.   That’s a huge change in the rankings, and it sets off a string of "secondary adjustments" in the polling; everybody else is pushed up or down 10 points or 20 or sometimes 30, based on a re-evaluation of previous polls with the new score for Trump.   A 30-point loss is only 3/10^th of one percent; it’s not actually a massive change; even Trump’s 96-point loss is a little less than one percent.   It just seems like a big number because, on most days, the secondary-effect movements are 3 points or less for almost everybody.  

            I think I am going to make some changes to the list process.  I think what I am going to do is:

1)      I’m going to add Stacey Abrams to the polling group, which will make 28 candidates.  I don’t really see that there’s a disadvantage to doing that; with daily polls people get tired of seeing the same names pop up again and again.  

2)     Tomorrow, I’m going to run a poll which is three random names and Stacey Abrams,

3)     Sunday I will do a poll of the four people who have been polled least often, whoever those are, and then

4)     Next week I will poll the candidates in order of their strength.   When we have 28 candidates each one would be polled once a week—4 times 7 is 28.   I think next week I’ll do the four strongest candidates against one another on Monday, the next four on Tuesday, the third heat on Wednesday, etc., getting down to the four weakest candidates (at the start of the week) being polled on the following Sunday.  

Then I’ll have to do something else the week after that, or I would wind up running the same polling groups again and again, thus not learning much.  Maybe I’ll go back to random groupings, but make the groups so that everybody is polled only once a week; I don’t know.   Anyway, thanks for participating/thanks for reading.