"Really wrong": These bad Fox News polls for Trump are oversampling Democrats by eight points, says ... Sean Hannity

Something fun from a few days ago via the Washington Examiner. Easy prediction: The cognitive dissonance experienced by Fox’s primetime hosts whenever their network drops a gruesome new poll on the president will eventually lead to some sort of internal push to let them conduct their own MAGA-friendly independent polling. That effort probably won’t succeed — imagine how humiliating it would be for the Fox News polling bureau to have its own anchors freelancing on election data — but the effort will be made. By the time we get to the thick of the race next summer, the imperative to protect the primetime audience from any news that’s discouraging for Trump, particularly polling-wise, will be overwhelming.

I assume Fox executives and Hannity will reach a compromise in which he doesn’t get to do his own surveys but also doesn’t have to cover any bad Fox News polls for Trump on the air either. In the meantime, he’ll casually suggest on his radio show that the current Fox polling is unreliable. Possibly even … fake news?

He’s half-right here, though:

“The New York Times–Siena College poll shows Trump in a dead heat in battleground states, matched up against the three leading Democratic candidates: Biden, Sanders and Warren,” Hannity said. “And I always say, and I believe, Trump doesn’t poll well. And I don’t know what’s up with the Fox poll. I look at their poll, I’m like, ‘OK, you’re oversampling Democrats by 8 points.’ I’m like — some outside company they hire — I’m like, ‘OK, they need new methodology because it’s really wrong.'”

I agree that the Times battleground poll is more important than Fox’s national polling and wrote about it this week. The Times poll is especially valuable because it sampled likely voters, the group that matters most on Election Day. Fox News’s latest national poll was terrible for Trump, showing him trailing Joe Biden by 12 points, but it was a poll of registered voters. Once you adjust the voter screen and focus in on the handful of states that will actually decide the election, Trump is in much better shape, probably not a lot worse than 50/50 even against a “strong” Democrat like Grandpa Joe.

But as for Fox supposedly oversampling Democrats by eight points, I don’t know where Hannity’s getting that. Presumably he’s looking at the fact that the sample in the last Fox poll was 49 percent Democratic and 41 percent Republican and imagining that an “accurate” poll would have equal numbers of D’s and R’s. Not true, and regular blog readers have encountered enough polls over time to know why. A good poll mirrors the demographics of the U.S. population and data collected over many years shows that the U.S. population almost always leans Democratic. The first graph here on Gallup’s page about this shows how the Dem advantage has waxed and waned since 2015. At last check, they enjoyed a five-point lead over Republicans — among all adults. When you look only at registered voters the gap narrows a bit, according to Pew:

Among the general public, recent Pew Research Center telephone surveys find that Democrats and Democratic-leaning independents outnumber Republicans and Republican leaners by about 7 percentage points, similar to what the 2018 General Social Survey found. This Democratic-Republican balance is one of the factors we use to weight Pew Research Center’s American Trends Panel (along with a core set of demographics including sex, race, age, region and other characteristics measured in high-quality government surveys). All of our survey reporting is based on weighted data, rather than raw data that has not been adjusted to meet these parameters.

Among registered voters, the Democratic advantage in party affiliation is typically about 3 percentage points smaller than it is among the general public in our surveys. The reasons for this are well understood: Compared with the general public, registered voters tend to be older, wealthier and more likely to be non-Hispanic whites and homeowners, all of which are characteristics associated with a higher probability of being a Republican.

So, typically, we’d expect a Democratic advantage of seven points or so among all adults and of four points or so among registereds. Hannity might be right, then, that Fox oversampled Democrats — a little, not by eight points. Certainly not by enough to account for the entirety of Biden’s 12-point lead in the poll.

What further complicates the picture is that Trump isn’t a “typical” political figure and at the moment we’re in political circumstances that are atypical even by Trump standards. How does impeachment affect the usual partisan split between Democrats and Republicans? Beats me. Could it be that left-leaning anti-Trump independents have begun to identify more as Democrats lately in support of Pelosi’s impeachment inquiry, boosting the share of Democrats in the total population? Is it possible that some Trump-wary Republicans have shifted into the independent column as the evidence of a quid pro quo mounts? Yeah, why not? According to Gallup, Democrats enjoyed an eight-point lead in partisan identification over Republicans among all adults as recently as the first quarter of this year. That lead shrank, then began growing again until it reached the five-point gap we see now. Who knows which direction it’ll go as the rest of the impeachment process plays out?

Another way to try to gauge whether Fox is oversampling Democrats is to look at the samples other major pollsters are seeing lately. They’re all working towards the same goal, getting the same proportions of key demographic groups in their samples as those groups enjoy in the general population — age, sex, race, education level, you name it. Then they ask whether each person identifies as a Democrat or Republican. Fox’s best attempt to mirror the general population produced a split of 49D/41R, or D+8. The latest poll from NBC and the Wall Street Journal had it 43D/35R (including leaners), or … D+8. The most recent survey from ABC and WaPo got 29D/23R (plus a whole lotta independents), or D+6. If you look at the very latest polling on the generic ballot, you find a range from D+6 to … D+8.

Again, doesn’t mean that won’t change tomorrow in Trump’s favor. And without a shadow of doubt, the margin is much tighter among likely voters and probably tighter still in the swing states. But there’s no need to resort to poll trutherism to explain an unhappy Fox News result.