For context, the president won Arizona in 2016 by 3.5%.  According to the AZ SOS website there are the following registered voters in Arizona:

Trump won the state by 3.5% and currently there are still 3.5% more registered Republicans than Democrats in AZ…

RealClearPolitics average polling for the U.S. Senate seat in Arizona has Democrat Kyrsten Sinema beating Republican combat Veteran Martha McSally by a little under one percent.  A close race one would say.

Some polls give more information than others.  The CNN poll admitted how many Democrats and Republicans they talked to, 25% Repub 25% Demon and 50% Other.  Those numbers do not mirror the above posted official AZ registration data or what the 2016 CNN exit polls say the turnout of our previous presidential (Trump) election was:  32% R, 28%D and 40% other, or the ACTUAL 2018 early voting turnout reported by the state of Arizona (see table below).  So, easily spotted, the CNN poll is wrong because (among other reasons) they are way under sampling Republicans by at least 7 points, but most likely by 10.

Other polls do not give you that information, but inferences can be made thanks to fake news CNN giving lots of details on their methods.

Regardless of the final results, a couple interesting things within those other polls:

• NYT Poll – A very weird statement cautioning readers that this is “just one poll”, like they are apologizing for having the evil Republican up by 2 points
• NYT Poll also shows what the results would be if the type of people that voted in 2016 voted again, +6 McSally.
• NBC poll – 44% have already cast their ballots, poll was taken starting on the 23rd (more than a week ago).  I would assume more than 50% have now voted early, which given the numbers, is great news for Republicans.

Back to the subject at hand…An election poll, any election poll, is based on what the pollster thinks voter turnout will be.  Obviously this time around, they are not basing turnout on what happened in 2016, actual voter registration numbers, or on 2018 early voting trends:

Early voting results are showing that Republicans are turning out at a HIGHER rate than in 2016 while Dems are slightly down:

• Republicans represent 43% of the early vote, they were 40% in 2016
• Democrats represent 33.5% of the early, the same as 2016

Given the early voting turnout, the polls showing Martha McSally losing (or tied) are either wrong (most likely) or in the next 5 days, including on election day, Republicans have to slow way down (like completely stop voting) and non-affiliated voters have to go from 24% of the turnout to 50%.  For example, to match the CNN polls turnout model, starting today the following would have to happen:

Obviously that wont happen, in fact since I started this post, more Republicans have mailed in ballots which means even more “other” voters need to turnout (and vote for the Arizona hating Democrat).

We do not know who any of these voters are actually casting ballots for, but the polls are based on a model that is different than the REAL turnout numbers, which means the polls are going to be wrong in Arizona.