For Whom the Polls Tell

<![CDATA[On Monday, I took a deep dive at a wave of polling out there, both national and in individual states, battleground and not, to help give you a better sense of where Campaign 2024 sits without all the spin and noise. It was among the longest columns I've written yet here at Hot Air, and it still didn't cover everything out there that raised my eyebrows just this week alone. So here is part two of the polling compendium. Full disclosure on my part - at the time Donald Trump was sewing up the nomination during the Republican primaries, I was very concerned that the former President's negatives, especially with suburban women and independents, were so great that barring multiple polling data showing me that he had a solid chance of winning, I didn't think November looked too rosy for the Republican Party, even with addled Joe Biden as his opponent. In the seven months since, and including a swap-out of candidates by the Democratic Party, I have seen a ton of data, not just the top line, but below the fold numbers that have convinced me that not only can Donald Trump win in November, like Nate Silver, he's still the odds-on favorite to win the Electoral College. Since Monday, two national polls joined were released by ABC/Ipsos and Fox News that pushed the Real Clear Politics average for Kamala Harris up by half a point - 1.5 to 2.0%. The ABC/Ipsos survey shows Harris with a four-point lead, while Fox News has the race at Harris +2. In 2020, Fox News' polling was ranked, in order of lowest to biggest prediction-to-actual error rate, in 13th place out of 20 polling outfits. Fox News missed 100% of the time in favor of the Democrats, and their average miss was 5.4%. Even if they improved their methodology enough to cut their error rate in half, Harris' lead is still not big enough to cover the 100% error rate over-sampling Democrats. To be fair in making historical comparisons, ABC and Ipsos were not partnered together in 2020. ABC's polling partner was the Washington Post, and Ipsos was married to Reuters. ABC/Ipsos currently has Harris ahead by 4, accounting for the half-point jump in the overall RCP average. But what's their track record? In 2020, Reuters/Ipsos had a miss rate of 5.1%, ranked 11th out of 20 polling groups, and when they missed in polling, missed 100% of the time in favor of the Democrats. ABC/Washington Post mirrored the Fox News polling record - 5.4% error rate, over-polling for Democrats 100% of the time.Early Thursday morning, another poll was released by the New York Times and Siena. Nationally, they have the race all tied up at 47. This polling is brand-new, and as of the writing of this column, had not been integrated into the RCP average. It will certainly bring Harris' lead back down to 1.9% again, which is where the average was before the debate. By the way, regarding the accuracy of NYT/Siena, in 2020, they missed by an average of 5.1% every time they polled, missing 92% of the time in favor of the Democrats. The only other poll out this week is the AtlasIntel poll I wrote about on Monday, showing Donald Trump with a 2.9% lead. AtlasIntel's overall miss rate in 2020, national and statewide, was 2.01%. Their final national poll between Joe Biden and Donald Trump predicted a 4.7% win for Biden. The final margin was 4.3%, so they over-polled for the Democrats, but just barely. They have Trump up over Harris just a tick under three points. If their accuracy this cycle matches past performance, Trump is ahead.  In the battlegrounds, there's a lot of material to digest since Monday, and we'll begin in the Peach Tree state of Georgia. The Atlanta Journal Constitution's poll came out, showing Donald Trump opening up a statistical lead once again over Kamala Harris, 47-44%. Why is this important? In mid-September, 2020, the same AJC poll showed the race between Trump and Biden in an absolute tie - 47-47%. The final margin? Biden beat Trump by .23% - 49.47-49.24%. In raw vote numbers, just a smidge over 11,000 votes separated the two. The AJC poll turned out to be pretty accurate six weeks out. Assuming they've got their finger on the same pulse and they have Trump up three, that's very good news indeed for the Republicans. In Pennsylvania, Donald Trump had reestablished a thin lead until mid-week, when Quinnipiac released a survey that suddenly sees Harris up 5. How much credibility are we to give this poll, based upon Quinnipiac's past performance? Not very much. In Real Clear Politics' polling scorecard for the 2020 campaign cycle, ranked from least to most average miss-to-actual result, Quinnipiac came in 19th out of 20 pollsters. Their average miss? A staggering 7.3%. When they missed, and it was often, they over-polled for Democrats. Only Monmouth had a consistently worse track record in that cycle than Quinnipiac. How bad was the Q poll in 2020? Take Maine's Senate race, for example. In mid-September of 2020, Quinnipiac all but shoveled the final dirt into the political grave of Senator Susan Collins. The Q poll reported that Collins' Democratic challenger, Sara Gideon, had opened up a 12-point lead over the incumbent Collins. A poll that far outside the margin of error had to mean certain doom for Collins, right? Collins not only was reelected, but won the race easily. The final margin was 9 points. Quinnipiac missed this race just six weeks out by a net 21 points. They have Kamala up 5 in the Keystone State? I'm not buying it. Why? Voter registration. Salena Zito has forgotten more about Pennsylvania politics than you and I will ever know. Her Washington Examiner column this week looks at Wilkes-Barre and the surrounding environs, and how the party affiliation and voter registration has shifted since the days of Barack Obama. In short, here is the problem the Democrats are running into in the rural parts of Pennsylvania - they're getting outworked in registering new voters, and the numbers are piling up. ]]>