Why and How I Very Publicly Predicted a Katie Hobbs Win in AZ.

There are reasons one wants to claim one’s capital from a successful thumb-nose-at-the-experts.  Most of these are self-serving in nature.  Punditry can make you look like a genius one day and Ralph Malph (Happy Days) the next.  I missed on several races, including the House.  I misjudged New York.  But I paid special attention to AZ.  One, we live here.

Two, Kari Lake tried to make her supporters believe the only way she could lose is by fraud.  But as I like to remind you, I keep receipts.  It is true an observer that is not deep into the minutae of politics might have found the victory of Governor-Elect Hobbs to be a surprise.  It really wasn’t.  But before I get into that, let’s talk about some clear indications that Lake knew she was in trouble.

The week before the election, Lake ran around Arizona dusting off old playbooks, like “Obamacare” and “elitism.”  She deliberately insulted John McCain supporters and played to the deepest darkest reaches of her base.  Any strategist will tell you that if you are not shoring up the middle at the end, you are probably behind.

The truth is, aside from partisan polling that bordered on the ridiculous, Katie Hobbs never trailed.  She was never up by ten points, nor down by eleven.  Both sides knew this.  Both sides know full well that her win was not only plausible, but if you paid attention to unbiased data, predictable.

While the count was going on, Lake’s team kept tweeting that Hobbs had no mathematical path to victory.  I will not comment on what they were thinking.  I honestly don’t care.  If we look at the exit polls, however, we can see that there was a glaring error that might have given them hope.

Edison Research reported Mark Kelly winning Maricopa County by by 51-45.  He indeed did win by more than six.  Curiously, it had Katie Hobbs losing Maricopa County by 49-50, as seen below-

Arizona Senate Exit Poll snippet

Arizona Governor Exit Poll snippet

The overall exit poll had Hobbs winning by one.  I knew one of these to be impossible.  There was no way Kelly could be only four points better statewide than Hobbs while winning Maricopa by seven points more.  Impossible.  Something was not going to line up.  It doesn’t mean the data itself was bad, it means it was insufficient.  In no way do I claim incompetence or malfeasance against Edison.

Data is data.  It happens.  When I tried to contact them to suggest they re-evaluate Maricopa, I got this response:


Nothing, even having identified myself as a member of the media.  So be it.  As of this writing the data still has not been recalculated to reflect the margin of slightly more than 2 points in favor of Hobbs in Maricopa.

But there are two things here that need to be discussed.  One, is that as I said before, the victory for Hobbs was predictable.  And one pollster nailed it.  Marist, the absolute 2022 gold standard of polling, came to this conclusion less than a week before the election.  To their credit, they refused to herd, or line up behind the conventional wisdom submitted by mostly right leaning pollsters.  This is what they found:


It is going to be hard to read but in that data is this nugget of goodness:  Katie Hobbs was winning Maricopa by 51-46.  The formula for a Democrat to win in Arizona is get to about 61 in Pima, 40 outstate, 51 Maricopa.  If this data was right, ballgame.  I could see Hobbs was winning more than double the crossover vote, which as you political vets might remember is where Ronald Reagan politically thrived, and basically, Lake was electoral toast.  The toplines don’t always matter, in fact, this is one of the weaknesses of 538’s model.

They make adjustments based on bias, though that is a tall order, and what they should really be doing is looking at the key elements of the internals.  Of course, this is referred to as unskewing.  Sometimes it is unskewing.  But sometimes, polls need to be unskewed.

In 2020 there was a poll released in Arizona that showed Joe Biden winning Maricopa by 12 percent and losing the state overall.  That is a poll that needs a good unskewing.  If a poll is skewed, such as for example, if Maricopa County was showing up as 40 percent of the likely vote instead of its usual 60, that needs to be addressed.  Topline data is only as good as the underlying accuracy.

But I digress.  Marist was right down the middle.  I knew it when I found the recalled vote:  Joe Biden 46, Donald Trump 45.  The 2020 electorate!  It matched up with my tracking of the race, and so I crunched some numbers.  Adjusting for some Republican base going home at the last minute I came up with Hobbs having 50.68 percent of the vote.  She would win 50.3.  (My bad, I will try to do better next time.)

As it turns out, I was off on my estimate because Lake actually pulled an extra point with Republicans I did not count on.

So I decided, since Lake went about the campaign making her election seem like an inevitability, to make my claim.  Some might have thought I was wishcasting but that was never the case.  (And for those that question my miss in the House, we were about two percentage points away from making that one a reality.)  I spent weeks on end crunching data, along with my other work, on specific races.

I only used non-partisan data in final projections, and was not a homer.  I correctly called Ted Budd’s win in N.C., for example.  I did this for one reason-in case Lake or Hamadeh or Cowboy Finchem or Masters wanted to go to court.  I had to establish a clear paper trail of data, projections, and evidence supporting them to show the world that not only was Lake’s victory not inevitable, she was not even favored.

Sure, we knew this would be a tossup all along.  But she wanted all of Arizona to believe she was going to win in a cakewalk, when the reality is it was close.  It was always close.  And my advice to anyone defending frivolous claims against Maricopa County is to subpoena Lake’s internals.

They know what it showed.  I could tell from her rhetoric.  Katie Hobbs was slightly favored throughout the campaign, and that never shifted.  I am very sure the Hobbs internals will show that as well.

Kari Lake, like Mehmet Oz, had a literal army of partisan polling swamp the aggregators in the final weeks, skewing the final predictions for those that did not pay attention.  But I paid attention.

I kept receipts.  Mark Kelly, Katie Hobbs, Adrian Fontes, and Kris Mayes won Arizona fair and square.

And yes, I want all of my work, if need be, to be used in court.  I want this post, my prior predictions, my reasoning, the internals, the projected R/D/I electorate by a local Democratic strategist to all be considered if this woman tries to contest a fair election.  So this is why I went public, and risked my punditry reputation, and put my brand on the line, so that there would be a track record of somebody getting it right.

So did I do this to help my brand?  Somewhat.  But what I was really trying to do was protect our democracy.

538 is only as accurate as the data it inputs, and people point to it as some kind of proof that there was funny business in Arizona.  This is wrong.

The election here was fair, well-managed, and predictable.  Printer certification claims are silliness, as any issues (and there were few) affected both parties equally.

The Republicans need to learn to lose with dignity as they watch Democrats win control of the electorate.


Do you think I deserve some support?  Would you like to join me and read unbiased, receipt-keeping political reporting?  Then click over to my right and join The Claw News.  I need to add to my Claw Family and I promise to keep my promises.

And my receipts.