A SCIENCE AND AN ART — It’s the quants versus the pundits all over again. The biggest prize in November’s elections is control of the 50-50 Senate. That much everyone can agree on. But Democrats’ chances of retaining the chamber in an otherwise challenging midterm differ significantly depending on whom you ask. On one side are the poll-based modelers, like FiveThirtyEight and Decision Desk HQ . As of this afternoon, FiveThirtyEight’s “Classic” polling model says Democrats have about a 3-in-4 chance to keep control of the Senate. In the other camp are the political handicappers or forecasters, including this correspondent, who largely see the battle to control the Senate as a jump-ball. (POLITICO’s Election Forecast, which are my predictions, calls it a “Toss-Up.”) Where can you see these discrepancies most clearly? There’s Pennsylvania, where Decision Desk HQ’s model gives Democrat John Fetterman a 70 percent chance to win, while FiveThirtyEight’s “Classic” model is even more bullish, at 76 percent. My view? I’ve had Pennsylvania in our “Toss-Up” column for months, and my competitors at the Cook Political Report recently downgraded Fetterman’s chances, moving the race from “Lean Democratic” to “Toss-Up.” Same with Georgia, which all the political forecasters call a “Toss-Up,” but where Democratic Sen. Raphael Warnock’s odds of winning are currently pegged at 74 percent and 66 percent in DDHQ and FiveThirtyEight’s polling-based models, respectively. Both sides are in agreement on Nevada, calling it a “Toss-Up,” but they differ on the other close races. FiveThirtyEight’s “Classic” model groups Nevada in with North Carolina and Ohio as races without a clear favorite, while forecasters consider the latter two as leaning toward Republicans. So what does the most famous political quant make of the difference? “I think the forecasters are very good at their jobs, though they also have a couple of predictable habits,” FiveThirtyEight’s Nate Silver told me, pointing to forecasters’ conservative approach to putting more races in a “Toss-Up” column and deliberate speed in making changes to account for new dynamics, like the Supreme Court’s June decision reversing federal abortion rights. “[Pennsylvania GOP candidate Mehmet] Oz has a shot, for instance,” Silver added, “but I just don’t think any data-driven answer would get you to Pennsylvania being a 50/50-ish race.” We forecasters don’t just rely on public data to make our predictions, of course. I talk to sources, occasionally hit the road on the campaign trail and — yes — pore over polling data. And it’s all of that mixed together that’s led me to the conclusion that neither candidate has a significant advantage in the race four weeks from today, even if Fetterman is currently leading Oz in the polls. Ultimately, Silver and the data modelers have one big thing right: The battle to control the Senate comes down to math. And a stalemate — the latest polls point to a 50-50 breakdown after the election, too — means Democrats would retain the chamber. Silver does see an important role for qualitative election forecasters like me, especially given polls’ consistent underestimation of Republicans in recent elections. That’s why he has another model — FiveThirtyEight calls it the “Deluxe” version — that includes “experts’ ratings.” Not surprisingly, Democrats’ odds are lower in that forecast. “This cycle, I’ve tended to trust the forecasters more than the polls on Wisconsin and Ohio, where they’ve had a more bearish outlook for Democrats,” Silver said. Ohio is a great example: FiveThirtyEight’s polling average has Democratic Rep. Tim Ryan ever-so-slightly ahead of Republican J.D. Vance, by 0.3 percentage points. But the site’s 2020 Ohio polling average had then-President Donald Trump ahead of Joe Biden by less than a point. Trump won by 8 points. Are the polls going to be wrong in the same direction, in the same places, as they were in 2020? It’s obviously too soon to say. But many qualitative forecasters are pricing in skepticism about the polls that’s missing in a model that relies solely or mostly on them. And that’s the big difference right now. Welcome to POLITICO Nightly. Reach out with news, tips and ideas at nightly@politico.com. Or contact tonight’s author at sshepard@politico.com or on Twitter at @POLITICO_Steve.
|
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.