I look at polling with a significant amount of skepticism.  No poll is right; most polls are wrong.  Some polls are wrong because of faulty methodology; others because they go into the survey process with a desired outcome and they make the numbers fit their narrative.  (Rasmussen and Trafalgar are in the latter category, in my opinion.)

My skepticism is grounded in my professional experience.  For nearly two decades, major portions of my jobs were spent analyzing numbers and the external factors influencing those numbers.   

I was an automotive “bluebook” editor for more than a decade.  The service I ran provided this information to the companies that you now use through the internet (although not the namesake of the type of publication, Kelley Bluebook, an entity which I can tell you more about than you’d care to know and which has its own, um, unique methodology of setting car prices) as well as to the manufacturers, dealers, insurance companies and financial institutions reliant on having accurate values of the cars they were financing or insuring, as well as companies that had macroeconomic needs like Jaguar, Volvo, GM and many other manufacturers. 

My individual area of specialty was European cars with a specific interest in post-war luxury, exotic and collectible cars.  For some vehicles–like popular models of BMWs and Mercedes of recent model years–we’d gather tens of thousands of data points from across the country that would point in a specific direction for determining current and residual values.  

For others, like 1960s Aston Martins, coach-built Bentleys, Porsche 356s and 1950s Volkswagen Beetles, data were sparse.  I had to look at historical data; current asking and selling prices; data origin; location of sale, anticipated national and global economic conditions, etc. to figure out “book value” and residual values.  

I don’t want to even tell you the headscratching that went into pricing something like a 1960s Ferrari 250 GT short wheelbase California Spyder, of which they manufactured only about 24, with (I believe) 23 known surviving vehicles.  Even seemingly non-related issues, like the death of Enzo Ferrari in 1988, rippled through the automotive market for years, impacting Ferraris, European luxury cars and (in my opinion) sending the US muscle car market on a decade-long roller coaster ride.  It demonstrated how something so non-economic as an individual’s death can shake markets.

Believe it or not, these skills integrated nicely into the healthcare industry, where I advised self-funded healthcare plans run by unions, government entities, and larger corporations (typically with 500 or more employees).  Planning those health care financials was not simply a matter of looking at the previous year’s totals and adding, say, five or nine percent.  You looked at the actuarial information about populations, but also things like the increasing or decreasing costs of specific treatments.  

For example, as CT scans became more common, the cost per service dropped.  In an employee population of a few hundred office workers, that might not make a big impact, but if you’re examining a regional Teamsters Health & Welfare Fund with tens of thousands of lives, that impacts hundreds of thousands of dollars in expense.  The introduction of a drug like Keytruda is now having a similar impact on cancer treatment costs for groups across the country, stabilizing costs and giving plans (and patients) more certainty.

Getting back on point:  I explain all this as a way of prefacing that I’m experienced in reviewing complicated numerical models, and that’s exactly what polls are.  In a lot of ways, polls are simpler than what I’ve done before because they don’t take outside influences into consideration (or at least, they shouldn’t).  Sometimes poll analysis is too simplified:  the frequent references to RealClearPolitics polling averages drive me nuts because it’s a simple arithmetic average and doesn’t take into account aspects of the component polls like sample size, sample make-up, and geographic distribution of the individual polls.  Add to that the obvious cherrypicking of which polls are used, and I find RCP to be more lucky than good when they get a race right.

RCP’s average gives equal weight to a poll of 600 Americans as it does a 2,500-person poll of so-called likely voters, for example.  In my mind, the former poll is essentially useless in forecasting the outcome of an election; the latter is far more reliable on the surface, regardless of minor flaws you may find as you dig deeper into the data.

I’m watching the trends in polls listed in RCP, FiveThirtyEight and other aggregators, and they’re all pointing to a shockingly good outcome for Democrats in the 2022 Congressional midterms.  Gerrymandering in Congressional districts will play a big part in the outcome but in general, from many analysis I’ve read, it seems to be a wash in the big picture:  nearly as many seats were secured for Democrats as Republicans thanks to post-2020 district redrawing.

Generally, generic Party preference polls are meaningless:  they’re not reflective of how any individual House or Senate race will turn out.  But what they do reflect is enthusiasm for the positions of the two major Parties.  And what they’re showing is simple:  voters are pushing against Republican overreach and authoritarianism, now favoring Democratic policies and management.

I’ll call it the Clarence Thomas effect:  After the Supreme Court struck down the Roe precedent, Republicans in states across the nation–particularly where they hold control of all branches of government–started to push through laws specifically attacking individual rights to things Americans came to take for granted:  access to abortion, availability of contraception, and the legality of LGBTQ marriage.  

These are overwhelmingly popular and accepted aspects of American life within the general public, and they’re the targets of Republican scorn.  Clarence Thomas told the whole of the US that these are things he and other Republicans want to dismantle in his concurring opinion to the Wade decision, and while members of the GOP poo-pooed the idea that Republicans would ever attack them, the American public remembered they made the same claim about Roe:  “Fool me once; shame on you.  Fool me twice…”

Another data point encouraging for Democrats:  turnout.  Since Trump’s anti-election fiasco started in November 2020, participation in elections has jumped.  Look, for example, at the Kansas anti-abortion ballot measure.  In a primary election, in Kansas, where basically only Republican nominations were up for grabs because most state office Democrats were running unopposed, a whopping 900,000 people turned out to vote.  That’s nearly as many as turned out for the 2018 general election with the governorship on the top of the ticket.  That. Is. Amazing.  

This started in the 2021 Senate runoffs in Georgia and continued on into the Virginia gubernatorial race in November 2021–a race won by a Republican, Glenn Youngkin, who was not a super-Trumper, but more of a Trumper by necessity to get the GOP nomination.  Youngkin’s proven to be more of a traditional Virginia Republican–more a traditional conservative than a radical–since he took office.

The last thing that I think will drive the midterms:  Trump fatigue.  Unless you’re a hardcore Trumpian, it’s really hard to maintain the level of anger the GOP needs to motivate their voters.  Republicans have themselves to blame for this.  They’ve spent the last seven years trying to make and keep everyone mad, particularly targeting Democrats.  But when push came to shove–the push being the January 6th domestic terrorist attack and the shove being overturning Roe–sane Republicans took a step back and said, “What the fuck are we doing?  This isn’t conservative.  This isn’t patriotism.  This isn’t America.”  

Yes, Trumpians won some key primaries in the past few months:  the GOP races for Pennsylvania governor, the Michigan 3rd Congressional, and a number of races in Arizona pop to mind.  But these are primaries, where only the most ardent supporters and political junkies vote.  Faced with the choice of radical Trumpians who tell people elections don’t matter versus a Democrat, I believe many Republicans will either hold their noses and vote Democratic, or they’ll simply leave those races blank, like they did in the 2020 Presidential race.  Remember in 2020, tens of thousands of ballots that were otherwise Party-line Republican votes left the Presidential ballot blank.  That’s likely to continue into these midterms.  (Incidentally, the same thing happened in 2016, only it was the Democrats who left the Presidential vote blank.  They won’t make the same mistake again.) 

I’m not going to temper my enthusiasm for a positive Democratic outcome in the midterms.  They’ll strengthen their hold on the Senate, and they’re likely to pick up a couple of governorships as well.  It also bodes well for lesser state offices.  

Concerning the House, I’m not convinced Democrats can maintain the majority, but I’m also no doom-and-gloomer.  It comes down to turnout:  Democrats need voters in non-safe R seats to overcome gerrymandering, voter suppression, and general Republican fuckery to win those seats.  The question is:  are there enough non-gerrymandered districts in play?  That, I do not know.  But if Georgia in 2021 Kansas this month are the templates, it can be done.  

Of course, there could be an “Enzo is Dead” moment at any time which throws everything into disarray, but at this point, Democrats need to concentrate on the difficult grunt work to secure victories where two years ago, they would have pre-conceded seats.  If they do, they’ll control both chambers of Congress and the White House.