A brief note about internal polls. And bullshit.

Liz Mair
5 min readSep 8, 2020

--

Several years ago, my best friend — who I will call “D” for the purposes of this piece — told me that I needed to write more stuff explaining stuff that was obvious to political practitioners to people who don’t do politics for a living, and who find it all very confusing.

I still feel like doing what D said involves me engaging in a level of pretentiousness and pseudo-intellectual snobbery that I’m not comfortable with as someone who basically gets paid to dig up dirt and dish it to reporters anonymously most of the day.

But I also recognize that people who have, you know, lives (what are they?), often times see things in the course of reading about politics and wonder “why is that?” or “is that really true?” and they would like an answer from someone who knows marginally more than them or is a practitioner.

Today is one such day, and so I’m writing the first of what may be many entries in which I attempt to do what D urged me to do.

Our topic is internal polls and how if a campaign is releasing them, odds are, they’re losing (or at least underperforming relative to their competitor).

The prompt for the column is this tweet from McClatchy’s Dave Catanese who is a good reporter, and whose writing you should read:

As Dave notes, the public polling shows a very different picture to what Team Trump is trying to paint, courtesy of its internals.

And it’s worth noting that non-public polling I have seen, conducted by people who are not running for office this year or committees invested in people running for office this year, shows more or less what the public polling shows with regard to these two states.

That’s a good initial clue that skepticism of what Team Trump is selling is warranted.

But there’s a more basic one: They’re showing you their internals. Or maybe even their “internals.”

I have no idea what the Trump campaign’s internals actually show, but it’s fairly irrelevant. The point is, they are very likely to be wrong, and they are also likely being released at this point in order to change a narrative, which is that Trump is getting his butt kicked in the Southwestern US. As ever, the campaign is probably hoping that by releasing their poll numbers, they can shift the narrative, which will then cause people currently leaning to Biden because they like a frontrunner, or they’re indecisive, or… whatever… to shift back to Trump and salvage the race for the President.

And that could happen. Their internals could also be right. But history does not bode well for Team Trump on this front, and it’s worth going through some of it to prove how this is so.

Back in June, we saw some great indicators of how the release of internal polling, unlike the release of the Kraken, is a great way to spot candidate weakness. Let’s start by looking at Kentucky.

Moderate Democratic Kentucky Senate candidate Amy McGrath was facing off against progressive Charles Booker. Booker, notably, was backed by lefty think tank Data for Progress

Ahead of election day, McGrath’s campaign put her ahead of progressive challenger Booker by 10 to 13 percent. Data for Progress had Booker up 8 percent over McGrath.

Clearly, one or both of these sets of internals had to be wrong, and were likely being released to create an advance narrative of winning, presumably in the vain hopes of making internal numbers turn into reality, like a magician can theoretically turn a shoe into a rabbit in the middle of a magic act. Or not.

Live shot of a campaign pollster

But both sets of internals were wrong. In the end, McGrath nabbed 45.4% and Booker nabbed 42.6%. The winning percentages themselves weren’t that far off (McGrath’s internals supposedly predicted she’d get about 42%, Data for Progress’ supposedly predicted Booker would get 44%). But the gap — i.e., what Trump’s team is now pointing to in Nevada and Arizona yelling “+6!!!” and “+5!!!”?

WRONG.

Released internals were also wrong — mega, mega wrong — in the New York primary that took place the same day.

New York Rep. Eliot Engel, who was delivered a shock defeat on the night, released his internals in the run-up to election day. The campaign tweeted that their polls showed Engel up by 8 points with a few days to go, and were supposedly holding to that number right up to election day.

But Engel lost by almost 15 points, taking just 40.6% to the 55.4% that went to progressive challenger Jamaal Bowman.

Elsewhere in New York, things did not go according to internal polling planning, either.

Suffolk County Legislator Bridget Fleming, who ran to unseat Rep. Lee Zeldin in Eastern Long Island, was also optimistic about her campaign based on internals — or at least tried to convince media and voters that she was.

An internal poll Fleming’s campaign released put her at 28.89 % to 28.83% for her opponent, Nancy Goroff, and 21.68% for another contender, Perry Gershon.

Guess who won on the night? Goroff, who won with 36.1%.

Guess who came second? Gershon, with 34.8%.

Fleming did come in at 27.6%, which I guess is close enough to 28.89% to count for… something. But the overall thrust of the poll — that she was winning — was absolutely false.

None of this is, or should be, a revelation. The possibility that it is, of course, is why I’m writing about it (thanks, D!).

Back in 2016, Libertarian presidential candidate Gary Johnson’s campaign shared internal polling showing an effective three-way tie in heavily-contested Utah with Johnson at 26 percent, Hillary Clinton at 27 percent and President Trump at 29 percent.

Johnson’s polling didn’t factor in candidate Evan McMullin, but the point still stands: Trump won Utah in 2016 with 45 percent, and Johnson managed 3.5 percent.

Dude.

So, this is the deal: If a campaign is releasing its internals, odds are that campaign is not in nearly as good of shape as they want you to believe. Take it from someone who has been tempted only about a million times — most prominently on Sen. Roy Blunt’s 2016 campaign — to release the internals to show everyone that dammit, they’re wrong, and we’re going to win… but also someone who has steadfastly refused to do that because it is a surefire tell of weakness (plus, it might constitute tempting fate). In 2016, no one outside the campaign and Missouri GOP infrastructure believed Blunt would win by 3-4 points, and it would have been lovely to show the world what Blunt team members knew. But the information was best kept to ourselves until election night when Blunt did indeed win by… 3 points.

So, let this all be a lesson. The next time a candidate hits you up for money, saying “our internals have us TIED”… no, they don’t.

The next time they say “but our internals have us LEADING”… maybe, but probably not by the margin they’re telling you.

Don’t believe the hype. Look at where the biggest concentration of public polls show a race.

They’re more likely to be accurate than numbers released in the political equivalent of a Hail Mary pass.

--

--

Liz Mair

I’m the Founder, President and Owner of Mair Strategies LLC, a boutique communications, public relations and opposition research firm.