The polsters always start polling registered voters-then switch the model by reporting likely voters

[samsingh]

this is what always shows the polls suddenly tightening for no reason. It would be interesting to report the two numbers throughout the campaign.

In a world where there are a lot of intelligent people, it's surprising how stupid things can be and that we accept.

[Fiendish Thingy]

1. And this year, LV polls are likely to miss most of the masses of new registrants. Nt

[TexasBushwhacker]

4. I was thinking the same thing

How would polls of "likely voters" include new registrants who have never voted before? How would they include formerly apathetic registered voters who are now "MAD AS HELL AND I'M NOT GOING TO TAKE THIS ANYMORE!!!" voters?

[Abnredleg]

5. Doubtful

New voters enter the electorate all the time so the pollsters deal with that by asking follow up questions for newly registered voters to determine likelihood of voting.

[Fiendish Thingy]

9. Not all pollsters screen LVs this way

Some define LVs as those who voted in previous, or multiple previous elections.

[Abnredleg]

10. Then new voters would by definition not be likely voters

which would be a major source of error. Hence the follow up questions.

[uponit7771]

11. The NYT poll only sampled 792 people mostly men IINM

[Abnredleg]

15. Correct

That’s a big sampling error that happens now and then. That’s why I don’t pay attention to individual polls but rather focus on the polling averages.

[pwb]

2. An all women poll would be nice eh?

Roevember !!

[brooklynite]

6. Every poll provides cross tabs that break out opinions by sex.

When the NYT reported that women’s political choices were shifting, everyone discounted the results.

[Indykatie]

8. Every Poll Doe NOT Provide Cross Tabs - That's Part of the Problem

Trafalgar is notorious for this. Cross tabs should be a requirement for polls released publicly. I saw one GOP sponsored poll that had the gender split as 47 % for women despite the fact that women have voted in larger numbers than men for every election. Of course this poll with only 47% women in their "likely voter" sample yielded big movement to the GOP.

[uponit7771]

12. RIGHT !?!? MSM, 538 & RCP should have the rule; no cross tabs no legitimacy

[mcar]

13. 95 women, MOE 10 points

[Hortensis]

3. Also, data for likely voters firm up as elections approach and then are on us.

Polls of likely voters themselves morph as the election approaches, with questions added such as, "have you sent for a mail-in ballot" and "have you already voted"? And to pick up newly registered voters, always "have you registered to vote"?

Pew has an informative article that discusses the "switch" from registered to likely. I suspect that since 2012 more voters have become involved earlier. Also, it discusses a presidential election, and this is midterms time.

Determining Who Is a ’Likely Voter’

For the first half of 2012, most polls on the 2012 campaign were based on registered voters. Now surveys are starting to base their horserace estimates on likely voters. Why the shift now, and how do you go about determining who really is more likely to vote?

A. Pew Research and many other polling organizations typically do not report on likely voters until September, after the nominating conventions have concluded and the campaign is fully underway. The reason many pollsters report only on registered voters in the period before then is that, while campaigns may seem to be in full swing earlier, most voters are not fully engaged in the election yet. And much of the hard work of mobilizing voters has not taken place and won’t occur until closer to the election.

Accordingly, any determination of who is a likely voter prior to September may contain a significant amount of error. ...

https://www.pewresearch.org/2012/08/29/ask-the-expert-determining-who-is-a-likely-voter/[/div

[karynnj]

7. I agree

Then any change you see is from the same place, same methology. Not to mention, in recent elections, the heuristic models to reflect who will vote have often not been good.

Couple this to pollsters observing that in both 2016 and 2020, the people sampled who actually responded were less Republican than their later estimates of the population. Naturally, they have worked to correct this. At some point, they likely will be overcompensating. Having a response rate that is less than 5 percent is the root of the problem.

[LonePirate]

14. Recent polls have not been kind to Dems so we better hope actual turnout resembles registered voters

Despite everything Biden and the Democratic controlled Congress have done since January 2021, the right has been very successful in framing this election about inflation and crime and not about rescinding rights or destroying our democracy. The average voter somehow thinks paying an extra $75/month for gasoline is a worse fate than living in a fascist nation with no rights.

[FBaggins]

16. Not really true

Some pollsters report LV throughout the cycle… some report adults/registered/likely and don’t switch.

Just as importantly… the switch to LV usually happens in mid September (for those pollsters that actually switch) - which was our strongest point of the cycle. So it doesn’t explain the recent change of apparent momentum.