Explaining the problem with polls

 Washington Examiner:

...

Ari Fleischer , the former White House press secretary for the Bush administration, is cautioning that a bulk of the midterm polls being eyed by analysts lean too heavily on registered voters instead of likely ones, thereby including troves of those who are unlikely to cast their vote.

"Registered voters don't vote. Voters vote. And so you need to switch to likely voters. Likely voters always vote more Republican than registered voters. And right now, the media is still using an overbroad pool of people called registered voters," Fleischer told Fox News's Laura Ingraham.

Republicans began the summer as the favorite to surf a red wave and win the midterm elections against the backdrop of Democrats being dogged by historical trends, unbridled inflation, soaring gas prices, lackluster approval ratings for President Joe Biden, and an uptick in crime.
...

However, Fleischer contends that underlying polling, like the ones used in many of those forecasts, also don't focus enough on battleground districts that will determine the balance of power in Washington, D.C.

"It's just an amazingly distorted way to present news to people who want to know what's the shape of November looking like. It doesn't matter what happens in the cities. It doesn't matter what happens in some of the more rural districts," Fleischer continued. "It matters in those battleground districts. That's where control of the House is at stake."

Fleischer is not alone in his concerns about polling. New York Times elections analyst Nate Cohn recently wrote a piece warning that there were similar patterns in this midterm spate of polling to prior cycles in which pollsters overestimated Democrats' prospects.
...

Fleisher is pointing out that likely voters in swing districts is what pollsters should be looking at.