What moves swing voters

[Josquius]

If the left needs to become identical to the far right to win then what's the point in winning?

So preaching a populist progressive message is now "identical to the far right"?

They're not doing that. That the right are having success pretending they are is the problem.

This is, again, exactly the fucking problem. You care more about posturing then winning.

I care about what is best for people. Not petty culture war bollocks.

You would rather scream about how racist and assholes blue collar workers are if they don't accept 100% of the left wing agenda then actually convince them to vote for us.

Holy reading failure batman. Didn't I just cover this?

We are not at all talking about giving up anything at all. Nobody in this thread has proposed giving up on anything like that.

Yet you have not given any answer for what you actually want.
Make no mistake. The people crying about how the moderate left are identical to the crazies won't rest until they are fully aligned with the far right.
Hell. Even then politics as football will find a way to keep them the enemy.

This is exactly the problem. The demand that unless you parrot exactly the issues as one element of the left demand with exactly their language and their emphasis, why, you must be in favor or "...giving up on fundamental human rights and pushing back several decades of progress that have made the world a better place..." Or perhaps you are a homophibic racist shitbag?

This is exactly the problem. The demand that unless you parrot exactly the issues as one element of the right demand with exactly their language and their emphasis, why, you must be in favor or "...giving up on fundamental human rights and pushing back several decades of progress that have made the world a better place..." Or perhaps you are woke?

[Sheilbh]

Sure - I don't use "gone to the GOP" as a moral condemnation. Just factually they voted Republican having previously voted Democrat. That's what makes them swing voters.

My point is that when we talk about swing voters we're always fighting the last war to an extent and we shouldn't go in with the idea of what that swing voter should look like - that all depends on your overall strategy, the coalition you're building and how you target it. It's not clear to me, for example, from what I've read that the Hispanic voters who have swung to the GOP in the last couple of years are necessarily working class voters (as defined in that poll). They might well be, but I don't know - as I say I think it's definitely relevant in 2008-16 but I think things have shifted.

In the last two big elections in 2018-20 the key swing voters have surely been college educated, white middle class voters because that's how the Democrats won. Are we sure they still won't be the key democgraphic and what's the message which works for them (I suspect that might not be populist progressivism)? Or is it the Hispanic voters as this new vulnerability for Democrats - what the message that works for them?

Obviously that goes for both sides - if the Democrats were going all in on populist progressive messaging then I imagine that Republicans would be working very hard to understand the message that undermines that or to focus on the least positive bits. So the other way of reading this is that if you're a Republican the key is to focus on the "woke moderate" position and hammer that every day, which I think they're doing pretty well.

The skill and the bit that moves from "one cool trick to win elections" to actually winning is obviously how you construct a message that: turns out your base, reaches undecided voters, convinces some of your opponent's voters to come over to you and isn't vulnerable to loads of attacks. That's a challenge and much as I'd love to think that populist left message is the answer I'm not sure it's that simple. That's also where I think the strategy/story point comes in because it needs to be more than a grab-bag of messages even if they poll or focus group well. There needs to be some coherence.

I read a piece (which I'll post in the Brexit thread because it's about the UK) which I think is possibly a little bit relevant to this.

[Berkut]

I am not one element of the right though. I am no element of the right.

I haven't given an answer for what *I* want?

I don't understand the question - aren't my political views on Languish pretty clear? Is there some question about that?

In the context of this particular discussion, what I want is to convince fellow progressives to focus more on winning elections, understanding why they are NOT winning elections, and fix that problem so that we can achieve our actual political goals. That is what I want.

But I get the feeling that isn't what everyone on the left wants. The immediate and visceral reaction to actual data and strategy around winning is pushed back against rabidly - to me, it is basically saying that you would rather be "right" then be in power and actually set things right.

I personally am opposed to much of what I consider to be "woke" left wing activism (and not, that isn't what the right defines as woke) on principled grounds, practical grounds, and in this narrow sense, political grounds.

[The Minsky Moment]

You seem to be working rather hard to wave this away.

I don't see why - I am in accord with your views on this and sympathetic to the point being made.  But I also agree with your point that the data should drive the analysis, not the other way around.  And the data is more equivocal than the NYT article suggests.

The NYT article, *explicitly* states that that the focus of the report is not exactly the same as the focus of the article, so your "To begin with..." implication that this is somehow some kind of discovery of some sort of nefarious bate and switch is bogus:

You can read the full poll results here. (If you do, note that the beginning of the report focuses on a Democratic-leaning group of working-class voters — who are relevant to primary elections — rather than the swing voters who have been my focus.)

The section you quoted - accurately in its original parentheses - appears at the very end of the article.  There is no indication before that point that the swing voter analysis was in fact only a small part of the overall report.  Any reasonable reader would conclude that the was the primary or even sole focus of the analysis unless and until they reached that last parenthetical sentence.  Even then, the NYT writer is suggesting that the non-swing voter portion is just something in the "beginning" of the report.  That is incredibly misleading because in reality the swing voter analysis takes up 2 pages in the middle of a 76 page report.  After that the report continues to discuss the full sample.  The messaging results of the swing voter analysis are never discussed again and the swing voter sub-sample is only mention three times in the next 30 pages, in each instance to note their preference for candidates that label themselves independent.

The NYT article simply gives a misleading impression about what the report was really about. 

And the actual poll results almost exactly align with the message of the NYT article:

I distinguish the numerical pull *results* from the "takeaway" analytical summary.  The reason I do so is that the takeaways -while not an unreasonable way of characterizing the results - do not give a proper sense of the relatively marginal nature of the effects and the large margin of error in the numbers.

And your claim that the error bars are so long that they all fall within the margin of error is rather disingenuous. They all fall within the margin of error if you overlap all five responses with one another. But the delta between the progressive populist response (the favorable response with this group) and the woke response (the least favorable) is significant, and only overlap in the very extreme. To the extent that you can dismiss the data on those grounds, you might as well simply throw out the entire poll, and in fact, the vast majority of polls that are trying, in good faith, to figure out nuance between competing positions.

All of the left messages overlap with progressive populist, and in the case of the bete noire woke progressive message, the overlap is quite substantial.  Moreover, the total distance from the median points is not great (around 54-47%).  Thus, the idea that these swing voters gush over prog pop and utterly despise woke prog is not supportable.  Over the collectivity there is mild preference for the former.

That distinction is significant for many reasons - to take just one it may be the case that you could change or flip those results by tinkering with the exact content of the message being polled.

[DGuller]

All of the left messages overlap with progressive populist, and in the case of the bete noire woke progressive message, the overlap is quite substantial.  Moreover, the total distance from the median points is not great (around 54-47%).  Thus, the idea that these swing voters gush over prog pop and utterly despise woke prog is not supportable.  Over the collectivity there is mild preference for the former.

Two bars overlapping does not imply "no statistical difference", confidence intervals can't be stacked additively like that.  If you want to know know the confidence interval for the difference, you have to calculate it directly (or know what the covariance term is).  You can't infer the confidence interval for the difference from two confidence intervals for the two means (and even if you assume independence, it's not as simple as checking for overlap, the standard deviations are not additive, variances are).

[PJL]

The real swing voter is one that doesn't normally vote at all. That's how populists get into power. See Weimar Germany's elections - the higher the turnout, the more votes Hitler got. It's the same with Brexit - highest vote share in any UK wide election/referendum etc for 20 years. Ditto in the 2020 US election - many voters came out for Trump to fight the Dems, but more voted against because of his record. Memories of the latter will fade, but the resolve of the former will remain, making him more electable in 2024.

IMO a high turnout is actually a bad sign for democracy, contrary to popular belief.

[The Minsky Moment]

Two bars overlapping does not imply "no statistical difference"

Good thing I didn't say that then.

"Thus, the idea that these swing voters gush over prog pop and utterly despise woke prog is not supportable.  Over the collectivity there is mild preference for the former."

[DGuller]

Two bars overlapping does not imply "no statistical difference"

Good thing I didn't say that then.

"Thus, the idea that these swing voters gush over prog pop and utterly despise woke prog is not supportable.  Over the collectivity there is mild preference for the former."

I was also addressing the original statement, even though I quoted the follow-up one:

However, it is important to note that the error bars are quite long and thus the all the apparent differences fall within the margin of error.  In particular there does not appear to be any material difference between the results for woke progressive, woke moderate, and mainstream moderate.  Progressive populism does seem to have a small edge but again within the margin of error.

You don't know whether the differences fall within the margin of error, you don't have enough information.  Something falling within the margin of error is synonymous with "not statistically significant".

[crazy canuck]

The real swing voter is one that doesn't normally vote at all. That's how populists get into power. See Weimar Germany's elections - the higher the turnout, the more votes Hitler got. It's the same with Brexit - highest vote share in any UK wide election/referendum etc for 20 years. Ditto in the 2020 US election - many voters came out for Trump to fight the Dems, but more voted against because of his record. Memories of the latter will fade, but the resolve of the former will remain, making him more electable in 2024.

IMO a high turnout is actually a bad sign for democracy, contrary to popular belief.

I think that is the part that is missing in this discussion.  If US democracy depends on attracting people who are willing to vote GOP at this time, God help the US.

[The Minsky Moment]

You don't know whether the differences fall within the margin of error, you don't have enough information.  Something falling within the margin of error is synonymous with "not statistically significant".

I avoided making statements about statistical significance.  I don't even know how the error margins were calculated or what standard was used. 

[The Minsky Moment]

The report itself does make a representation about statistical significance:

"Similarly, if the bars around one dot overlap with the bars around another
dot, we cannot conclude that there is any statistical difference between the
two characteristics."

[DGuller]

You don't know whether the differences fall within the margin of error, you don't have enough information.  Something falling within the margin of error is synonymous with "not statistically significant".

I avoided making statements about statistical significance.  I don't even know how the error margins were calculated or what standard was used.

As I already said, it's the same thing.  Margin of error and statistical significance are two terms for exactly the same concept, both require a standard to be picked.  Something that's within the margin of error is not statistically significant, and vice versa.  When you say that something is within the margin of error, you're saying that it's statistically insignificant, regardless of whether you think you're avoiding saying that or not. 

Such a statement, regardless of which of the two equivalent phrases you use, are meant to dismiss a data-driven argument.  I'm being pedantic about it because your dismissal was unwarranted.

[DGuller]

The report itself does make a representation about statistical significance:

"Similarly, if the bars around one dot overlap with the bars around another
dot, we cannot conclude that there is any statistical difference between the
two characteristics."

They're wrong about that then.  It's a common mistake.

[The Minsky Moment]

If the authors of the report are denying statistical significance and given the report does not otherwise disclose the method of calculation or standard uses, I would be more hesitant than you to ascribe significance that the authors do not.

[DGuller]

If the authors of the report are denying statistical significance and given the report does not otherwise disclose the method of calculation or standard uses, I would be more hesitant than you to ascribe significance that the authors do not.

The good thing about math is that 2+2=4 even if Gauss for some reason wrote that it's 5.  The math is what it is, regardless of what people with even very high authority say it is.  What the authors say based on fallacious logic should have no bearing on what you think is correct.

I do doubt the underlying data more now than before you pointed out this paragraph, since if the authors get the basic math about confidence intervals wrong, then it's harder to trust them to get the softer aspects of statistics like sampling bias right.  My hesitancy would come from that part, not because I would trust them that 2+2=5 because it's useful for my argument.