Sanders against Trump vs. Biden against Trump

[SingingAnalyst]

Is there any state (besides Vermont) where Sanders does better against Trump then Biden does against Trump? (Significantly better, as in more than 1 percentage point).

Thanks.

[Former Dean Phillips Supporters for Haley (I guess???!?) 👁️]

Is there any state (besides Vermont) where Sanders does better against Trump then Biden does against Trump? (Significantly better, as in more than 1 percentage point).

Thanks.

It is likely to be either most states or not many states, and there is no real way to tell which of the two it is for sure at this time.

[SaneDemocrat]

Colorado, Hawaii,Indiana,Iowa, Kansas, Kentucky, Maine, Mass, Michigan. Minnesota,NH Vermont Wisconsin, West Virginia

[Troll and Sock Puppet. Delete my ACCOUNT!]

Colorado, Hawaii, Indiana,Iowa, Kansas, Kentucky, Maine, Mass, Michigan. Minnesota,NH Vermont Wisconsin, West Virginia

[SingingAnalyst]

Michigan? Even though a Glengariff poll of 600 showed Biden up by 7, but Sanders only up by 4?

Michigan is one state I might think would be in this category, but for this poll. Are there really many people in Kent, Livingston, Oakland, and Wayne Counties who would vote for Sanders, but not Biden, against Trump in November?

[Bidenworth2020]

Probably Hawaii, New Mexico, and maybe Alaska.

[Former Dean Phillips Supporters for Haley (I guess???!?) 👁️]

Michigan? Even though a Glengariff poll of 600 showed Biden up by 7, but Sanders only up by 4?

Michigan is one state I might think would be in this category, but for this poll. Are there really many people in Kent, Livingston, Oakland, and Wayne Counties who would vote for Sanders, but not Biden, against Trump in November?

I will give a mathematical example, which should be right up your alley given your username. It is a simplified example so that the math doesn't get overly complicated, using hypothetical poll #s to illustrate the point that Sanders could well be more "electable" and actually do better than Biden even if Biden polls 3 points better than Bernie and even if a poll having him doing 3 points better is completely accurate.

The point is not necessarily that Sanders is more electable (maybe he is, maybe he isn't, my point is not to take any firm position on that), the point is just that you can't conclude too much from these poll comparisons, because they ignore some crucial factors (namely the influence of turnout).



Let's say a poll has Biden +1 and Bernie -2 against Trump (so Biden is nominally 3 points stronger in terms of electability).

Underneath the hood, you have the following #s leading to these toplines:


Biden:
wins young voters 70-30%
loses old voters 42%-58%

Bernie:
wins young voters 75%-25%
loses old voters 38%-62%

However, the crucial assumption made by the poll is that in both cases, regardless of who the candidate is, that the same electorate will turn out to vote. In both cases, the assumption is that 70% of the electorate will be made up by old voters, and 30% by young voters.

If you multiply out those numbers by the 70-30 vote share (old-young) split, you end up with:

Biden 50.4% - Trump 49.6% (Biden ~ +1)

and Trump 50.9% - Bernie 49.1% (Trump ~ +2)



However, in reality the vote share of young vs old voters is not invariant to who the candidates are. If Biden is the nominee, it is plausible that young voters (and also some other groups not taken into account in this simplified example) may be less enthusiastic and less likely to vote than they would be if Bernie were the nominee.

For example, suppose that with Biden, young voter turnout were less, such that 72% of the electorate were old and 28% young, whereas with Bernie, young voter turnout were higher, such that 67% of the electorate were old and 33% young.


Multiply out the support #s by the turnout, and you get the following:


Biden:

70% support from young voters multiplied by 28% vote share + 42% support from old voters multiplied by 72% vote share = 49.84% total support (Trump gets the remaining 50.16%).


Bernie:

75% support from young voters multiplied by 33% vote share + 38% support from old voters multiplied by 67% vote share = 50.21% support (Trump gets the remaining 49.79%).


So in this case case, despite the poll being entirely accurate in terms of the % support that both Bernie and Biden get, and despite Biden doing 3 points better than Bernie in the poll's matchup against Trump, in reality Biden ends up losing and Bernie ends up winning because of turnout differences that are not taken into account by the poll, since the poll (incorrectly) questions the same voters regardless of who is the candidate, thereby assuming away any variation in turnout/enthusiasm between candidates.

[SingingAnalyst]

Michigan? Even though a Glengariff poll of 600 showed Biden up by 7, but Sanders only up by 4?

Michigan is one state I might think would be in this category, but for this poll. Are there really many people in Kent, Livingston, Oakland, and Wayne Counties who would vote for Sanders, but not Biden, against Trump in November?

I will give a mathematical example, which should be right up your alley given your username. It is a simplified example so that the math doesn't get overly complicated, using hypothetical poll #s to illustrate the point that Sanders could well be more "electable" and actually do better than Biden even if Biden polls 3 points better than Bernie and even if a poll having him doing 3 points better is completely accurate.

The point is not necessarily that Sanders is more electable (maybe he is, maybe he isn't, my point is not to take any firm position on that), the point is just that you can't conclude too much from these poll comparisons, because they ignore some crucial factors (namely the influence of turnout).



Let's say a poll has Biden +1 and Bernie -2 against Trump (so Biden is nominally 3 points stronger in terms of electability).

Underneath the hood, you have the following #s leading to these toplines:


Biden:
wins young voters 70-30%
loses old voters 42%-58%

Bernie:
wins young voters 75%-25%
loses old voters 38%-62%

However, the crucial assumption made by the poll is that in both cases, regardless of who the candidate is, that the same electorate will turn out to vote. In both cases, the assumption is that 70% of the electorate will be made up by old voters, and 30% by young voters.

If you multiply out those numbers by the 70-30 vote share (old-young) split, you end up with:

Biden 50.4% - Trump 49.6% (Biden ~ +1)

and Trump 50.9% - Bernie 49.1% (Trump ~ +2)



However, in reality the vote share of young vs old voters is not invariant to who the candidates are. If Biden is the nominee, it is plausible that young voters (and also some other groups not taken into account in this simplified example) may be less enthusiastic and less likely to vote than they would be if Bernie were the nominee.

For example, suppose that with Biden, young voter turnout were less, such that 72% of the electorate were old and 28% young, whereas with Bernie, young voter turnout were higher, such that 67% of the electorate were old and 33% young.


Multiply out the support #s by the turnout, and you get the following:


Biden:

70% support from young voters multiplied by 28% vote share + 42% support from old voters multiplied by 72% vote share = 49.84% total support (Trump gets the remaining 50.16%).


Bernie:

75% support from young voters multiplied by 33% vote share + 38% support from old voters multiplied by 67% vote share = 50.21% support (Trump gets the remaining 49.79%).


So in this case case, despite the poll being entirely accurate in terms of the % support that both Bernie and Biden get, and despite Biden doing 3 points better than Bernie in the poll's matchup against Trump, in reality Biden ends up losing and Bernie ends up winning because of turnout differences that are not taken into account by the poll, since the poll (incorrectly) questions the same voters regardless of who is the candidate, thereby assuming away any variation in turnout/enthusiasm between candidates.

Thank you. That is an excellent argument (it even serves as an example of Simpson's Rule), and you have offered perhaps the most persuasive argument I have seen as to how Sanders could be more electable. I will certainly consider this between now and March 10th.

[Politician]

Sanders does better in the green states. Unsure about OH, IL, and AZ.

[Progressive Pessimist]

Sanders does better in the green states. Unsure about OH, IL, and AZ.

That very well could be a possible primary result map.

[Skill and Chance]

Sanders does better in the green states. Unsure about OH, IL, and AZ.

Seems right to me.  I'd be inclined to flip Ohio, but it was a big enough Trump win last time that it won't matter for the outcome.

Edit: I also don't think Sanders does better than Biden in California.  Way too many high income Dems for that.

[Amenhotep Bakari-Sellers]

Yes, WI, MI, PA, were lost last time with defections to Gary Johnson, Biden may cause defections by Bernie, like last time to Chafee, otherwise, it's a 279 map

[Xing]

We’re never going to know for sure, since we’ll never get a direct comparison, and the overall strength of the campaign, as well as the national environment will be much more important factors than how good of a “fit” each candidate is for a state, but I imagine all things being equal, Sanders might do a bit better than Biden in New England, the West, and the Upper Midwest. Basically his strong regions in the primary.

[Skill and Chance]

What's tricky about this is that the range is so much wider with Sanders.  Anything between reverse 1980 and reverse 1996 (probably including a strongish independent campaign either way) is possible with Sanders, while Biden likely has a much tighter range from perhaps D+5 to R+0.5 in the PV with a near uniform swing from the 2016 likely in the EC give or take a GA or ME.  It's also pretty clear Biden has to win the PV by even more than Clinton to win the election, while Sanders could win with a substantially closer PV margin.  An EV/PV split in his favor isn't even out of the question.

I think Trump is more likely to break out and have a landslide win than Sanders, but if the election is close, I expect Sanders to win.  Even though he is on paper stronger, Trump should probably prefer Biden as an opponent because a race against Sanders is so much less predictable. 

I will say Trump winning with another EV/PV split against Biden basically ensures AOC is the 2024 nominee, while a Sanders loss basically ends the Squad's influence over the party.

[Senator-elect Spark]

Better
Worse

[😥]

Better
Worse

[It’s so Joever]

Just my hunches since polling this early out is sh**t.
Better: (By more than 2% margin)
Vermont (Sorry had to)
Maine
South Dakota
Iowa
Oregon

Better (By 1-2%)
New Hampshire
Wisconsin
Washington

Same-ish
Rhode Island
Tennessee
Kentucky
West Virginia
Indiana
Michigan
North Dakota
Missouri
Colorado
Wyoming
Minnesota

Worse (1-2%)
Massachusetts
New York
Ohio
Arkansas
Idaho
New Mexico
Alabama
Mississippi

Worse (More than 2%)
Connecticut
New Jersey
Pennsylvania
Maryland
Delaware
Virginia
North Carolina
South Carolina
Georgia
Florida
Louisiana
Illinois
Nebraska
Kansas
Oklahoma
Texas
Utah
Nevada
California
Arizona
Alaska

[SingingAnalyst]

Michigan? Even though a Glengariff poll of 600 showed Biden up by 7, but Sanders only up by 4?

Michigan is one state I might think would be in this category, but for this poll. Are there really many people in Kent, Livingston, Oakland, and Wayne Counties who would vote for Sanders, but not Biden, against Trump in November?

I will give a mathematical example, which should be right up your alley given your username. It is a simplified example so that the math doesn't get overly complicated, using hypothetical poll #s to illustrate the point that Sanders could well be more "electable" and actually do better than Biden even if Biden polls 3 points better than Bernie and even if a poll having him doing 3 points better is completely accurate.

The point is not necessarily that Sanders is more electable (maybe he is, maybe he isn't, my point is not to take any firm position on that), the point is just that you can't conclude too much from these poll comparisons, because they ignore some crucial factors (namely the influence of turnout).



Let's say a poll has Biden +1 and Bernie -2 against Trump (so Biden is nominally 3 points stronger in terms of electability).

Underneath the hood, you have the following #s leading to these toplines:


Biden:
wins young voters 70-30%
loses old voters 42%-58%

Bernie:
wins young voters 75%-25%
loses old voters 38%-62%

However, the crucial assumption made by the poll is that in both cases, regardless of who the candidate is, that the same electorate will turn out to vote. In both cases, the assumption is that 70% of the electorate will be made up by old voters, and 30% by young voters.

If you multiply out those numbers by the 70-30 vote share (old-young) split, you end up with:

Biden 50.4% - Trump 49.6% (Biden ~ +1)

and Trump 50.9% - Bernie 49.1% (Trump ~ +2)



However, in reality the vote share of young vs old voters is not invariant to who the candidates are. If Biden is the nominee, it is plausible that young voters (and also some other groups not taken into account in this simplified example) may be less enthusiastic and less likely to vote than they would be if Bernie were the nominee.

For example, suppose that with Biden, young voter turnout were less, such that 72% of the electorate were old and 28% young, whereas with Bernie, young voter turnout were higher, such that 67% of the electorate were old and 33% young.


Multiply out the support #s by the turnout, and you get the following:


Biden:

70% support from young voters multiplied by 28% vote share + 42% support from old voters multiplied by 72% vote share = 49.84% total support (Trump gets the remaining 50.16%).


Bernie:

75% support from young voters multiplied by 33% vote share + 38% support from old voters multiplied by 67% vote share = 50.21% support (Trump gets the remaining 49.79%).


So in this case case, despite the poll being entirely accurate in terms of the % support that both Bernie and Biden get, and despite Biden doing 3 points better than Bernie in the poll's matchup against Trump, in reality Biden ends up losing and Bernie ends up winning because of turnout differences that are not taken into account by the poll, since the poll (incorrectly) questions the same voters regardless of who is the candidate, thereby assuming away any variation in turnout/enthusiasm between candidates.

Again, excellent analysis. Now that I have thought about it for a couple of days (and, frankly, spoken to a Biden supporter who nonetheless says she would vote for any Democrat against Trump), I have a couple of responses.

First, you have the election extremely close in either case, so the slightest variable (weather, comment by a downballot candidate, etc.) could flip either outcome the other way.

Second, it appears you are assuming no significant difference in "older voter" turnout with Sanders as compared to Biden, right? Me too. So, with the number of older voters as a baseline (and taking your numbers of a 42-58 split with Biden and a 38-62 split with Sanders), we have...

...in Biden's case, 0.39 "young" voters for every older voter, with 0.273 for Biden and 0.117 for Trump.

...in Sanders' case, 0.49 "young" voters for every older voter, with 0.3675 for Sanders and 0.1225 for Trump.

Thus, with Sanders instead of Biden, you are assuming a net increase of "young" voters equal to 10% of older voters-- with 94.5% of them voting for Sanders. What if a few of them decide to vote Trump instead-- maybe 90-10? That alone would swing the election based on the tight race you have.

Third, with "young" voters, it is a matter of them voting for Sanders vs. staying home or voting third party. "Older" voters (including many in heavily Cuban, swing-state Florida) may be more likely to be turned off by the "socialist" label, but instead of voting 3rd party or staying home, they'll vote Trump. So, for your theory to work, there would need to be at least twice as many "young" voters switching from nonvoter or 3rd-party voter to Sanders as there are "older" voters switching from Biden to Trump.

This absentee voter is still persuadable either way, and probably won't make up his mind until he's practically in the metaphorical voting booth.

[Forward]

I think Biden would be the overall stronger candidate. Bernie would win over working class whites, but might struggle with affluent whites and to a lesser extent blacks, while Biden has widespread appeal to all of these groups.