The Michigan primary is the best evidence yet that Joe Biden is in trouble
2024 is the first presidential election cycle since 2008 that I have not had a particularly strong intuition for who the general election winner would be. In every cycle between 2008 and 2020, my basic intuition was proven correct. This could be controversial at times, such as when I made the case for Trump’s likely over-performance in 2016. But in a strange way, my view of Biden’s likely over-performance in 2020 was even more controversial. Biden never had a vocal online media fanbase in the 2020 Democratic primaries, so online commentators of both Left and Right would routinely dismiss the prospect of him beating Trump — often scorning the very idea as a foolish impossibility.
Now, for the first time in the 2024 cycle, I’m beginning to develop an intuition. That’s because the same data which inclined me toward a favorable outlook on Biden’s general election prospects in 2020 have begun to incline me toward the opposite in 2024 — an unfavorable outlook on Biden’s prospects, and by extension a favorable outlook on Trump’s. The Michigan primary results yesterday contain some of the most illustrative data yet for why this is so.
A lot of people didn’t want to hear it at the time, but Biden’s performance in the 2020 Democratic primaries augured his subsequent victory in the general election. Those who didn’t want to hear this at the time included obstinate Bernie Sanders supporters, who furiously argued that Biden couldn’t possibly win against Trump, as well as obstinate Trump supporters, who had likewise convinced themselves that Biden obviously had no shot.
Vanishingly few online “influencers” supported Biden in the 2020 Democratic primaries — thus he had a drastically quieter online cheering section than Sanders, Elizabeth Warren, and other Democratic figures who inspired more rabid followings. Biden’s most reliable primary constituency, older Black voters in the South, were not well-represented on social media. Tweets, articles, and videos highlighting Biden’s electoral viability were conspicuously unpopular online, as I personally became aware. This pattern created a strong algorithmic incentive to deny Biden’s electoral viability, at least if you wanted to maintain or increase your online engagement. Since that’s what people often want to do (maintain or increase their online engagement) there was a huge blindspot that emerged in both Left and Right-wing online media about Biden’s electoral viability. The dynamic evolved somewhat when Biden officially became the Democratic nominee, and consequently the last man standing against Fascist Nazi Hitler Trump: Left-liberal media naturally consolidated, desperate to defeat Trump at any price. But during the primaries at least, Biden’s political strength was widely discounted by cross-ideological online media. The confusion was compounded by people in general having a chronic inability to distinguish between descriptive analysis and normative analysis — that is, if you say that Biden is electorally viable, you’re not necessarily saying that Biden is good, or that it’s good Biden is electorally viable. All you’re trying to do is empirically illustrate the electoral viability of Biden. Whether this viability is a good or bad thing politically, morally, geo-strategically or otherwise is a separate question entirely.
What most suggested to me that Biden stood a favorable chance of winning the general election against Trump was Biden and Democrats’ consistent over-performance in the 2020 presidential primaries. Particularly in February through March of 2020 — before the widespread introduction of COVID-related measures to expand mail voting. (Just in case anyone’s about to write an angry comment complaining about my shameful inattention to their election fraud theories, which I have already examined as thoroughly as could be reasonably expected, and rejected.)
It wasn’t just Michigan where Biden over-performed in the 2020 primaries, though his showing there carried extra political weight given the status of Michigan as a vital Midwestern “swing state.” Trump was the first Republican to win Michigan since 1988, and Biden’s core argument in the 2020 primaries was that he was the safest bet for Democrats to win it back and save America from a Democracy-Destroying™ second term of Trump. This argument — however dopey — eventually did prevail, as could have been foretold by trends that became discernible in February 2020.
Presidential primaries are underrated for their predictive power of subsequent general election results, as I’ll briefly try to demonstrate in this post. If the predictive power outlined here continues through 2024, it portends very poorly for the general election prospects of Biden — and, likewise, portends well for Trump.
Biden received about 217,000 fewer votes in the Michigan Democratic primary yesterday than he did in the 2020 primary — an approximately 26% decline in the number of Democratic primary votes received. What might this tell us about Biden’s prospective general election performance in 2024?
The first thing to look at is the past Michigan primary performances of Barack Obama, since Obama in 2012 was also an incumbent Democratic president running for re-election. Inconveniently, Obama happened not to be on the Michigan primary ballot in 2008 due to a technical dispute over delegate allocation, which at the time seemed biblically important — at least to operatives, bloggers, and myriad political obsessives (myself included) — but today seems so unimportant as to not even warrant a quick summary. The nub is that “Uncommitted” got 40% of the vote in 2008, largely as a proxy for Obama, whose name was not on the Michigan primary ballot.
So, using the 2008 “Uncommitted” vote tally as an imperfect proxy for Obama, we can tentatively say he received 250,000* votes in the 2008 Michigan Democratic primary. (If he had been on the primary ballot, the number of votes he received would’ve almost certainly been considerably higher.) Obama went on to win the 2008 general election, then ran for re-election in 2012, receiving 174,054 votes in the 2012 Michigan Democratic primary — virtually unchallenged as the incumbent president. Going by raw numbers, there was *22% decline in the number of Michigan Democratic primary votes received by Obama between 2008 and 2012:
Obama votes, Michigan Democratic primary
2008: 250,000*
2012: 194,887 (-22%)
*As discussed above, Obama was not on the 2008 primary ballot in Michigan for unimportant technical reasons. “Uncommitted,” largely functioning as a proxy for Obama, received 238,168 total votes and 40% of the overall vote. So, I’m revising upward to 250,000 as a rough estimate of the number of votes Obama could have likely received if he appeared on the ballot as normal. (It’s probably an underestimate. For reference, Hillary Clinton — the only major Democratic candidate on the 2008 ballot in Michigan — received 328,309 votes.)
The trend in Obama’s primary election performances foretold subsequent trends in his general election performance. Obama won Michigan twice, in 2008 and 2012, but in the 2012 general election he received 303,111 fewer votes than he had in 2008. Obama won Michigan by 16 percentage points in 2008, so to only win by 9 percentage points in 2012 was a marked reduction in both marginal (-7%) and absolute votes received (-11%) — even though Obama was still able to win Michigan comfortably overall. A key difference between Obama then and Biden now is that Obama was starting his 2012 re-election campaign with a much higher floor — a giant 16 point win in 2008 — while Biden won Michigan by less than 3 points in 2020. Hence, Obama was in a position in February 2012 where he could afford to lose a few percentage points, or maybe even more than a few, in his marginal advantage. Biden in February 2024 has nowhere near as much room for error.
Total votes cast in the Michigan Democratic primary decreased by 67% between 2008 and 2012 — correlatively, Obama would go on to receive 11% fewer votes in Michigan in the 2012 general election, with his margin of victory 7 points smaller than it was in 2008. If Biden underwent the same marginal shift in 2024, he would lose Michigan to Trump in November by 4 points. And depending how you look at it, Biden is already atrophying support in Michigan at greater rate than Obama did between 2008 and 2012. In the 2020 Democratic primary, Biden received about 40% more votes than the winner of the previous Michigan primary, Bernie Sanders, who narrowly beat Hillary Clinton in 2016. And now in the 2024 primary, Biden has received about 26% fewer Democratic votes than he did in 2020. In the 2012 primary, Obama received about 22% fewer votes than he did in 2008.
Total votes cast for the eventual nominee, Michigan Democratic presidential primary:
2008 (Obama): 250,000* (--)
2012 (Obama): 194,887 (-22%)
2016 (H. Clinton): 581,775 (+198%)
2020 (Biden): 840,360 (+45%)
2024 (Biden*): 623,415 (-26%)
*Estimate, see above
*Preliminary
These numbers looked good for Biden in 2020, and now look bad for Biden in 2024. Biden is atrophying a greater percentage of Democratic primary votes (-26%) than Obama was atrophying in 2012 (-22%) and as noted above, Biden has far less electoral breathing room than Obama did.
Biden looks to be on the opposite trajectory as he was in 2020, trending negative in his Michigan primary vote total (-27%) while Trump trends positive (+18%). The same was true earlier this month in South Carolina. Biden received 54% fewer votes in the 2024 South Carolina Democratic primary than he did in the 2020 primary, and overall Democratic primary turnout was down 76% compared to 2020. The South Carolina GOP held no presidential primary election in 2020, instead awarding all delegates to Trump by default, so there is no direct point of comparison to Biden four years ago — but relevant trend-lines may still be discerned:
Total votes cast, South Carolina Republican presidential primary:
2004: No primary held
2008: 445,499 (--)
2012: 603,770 (+36%)
2016: 740,881 (+23%)
2020: No primary held
2024: 755,800* (+2% from 2016)
*Preliminary
South Carolina is of course nowhere near as important as Michigan for the purposes of modern Electoral College calculations, but whichever demographics are fueling Trump’s upward trajectory in South Carolina can be expected to be extrapolatable to other states.
Total votes cast, South Carolina Democratic presidential primary:
2004: 293,843 (--)
2008: 532,151 (+81%)
2012: No primary held
2016: 272,379 (-49% from 2008)
2020: 539,263 (+98%)
2024: 131,472* (-76%)
*Preliminary
As we can see, Democratic primary turnout has trended substantially downward in both South Carolina and Michigan — the opposite as in 2020, when the same trends pointed to a Biden general election advantage over Trump. If the correlation holds in 2024, the advantage goes to Trump. Republican primary turnout is trending upward — not unlike what happened in the 2016 primaries, which augured Trump’s ultimate general election victory.
Total votes cast for the eventual nominee, Michigan Republican presidential primary
2008 (McCain): 257,985 (--)
2012 (Romney): 409,522 (+59%)
2016 (Trump): 483,753 (+18%)
2020 (Trump): 640,522 (+32%)
2024 (Trump*): 756,851 (+18%)
*Preliminary
It should be noted that Biden received hundreds of thousands more votes in the 2020 Michigan Democratic primary than Hillary Clinton in the 2016 primary, which she lost narrowly Bernie Sanders. Biden also got substantially more votes than Trump did in the (effectively uncontested) 2020 Michigan Republican primary. This did not look good for Trump at the time, because although he gained in total primary votes between 2016 and 2020 — not always the case for incumbents, as seen with Obama and Biden — the increase was easily dwarfed by a larger concurrent increase in Democratic primary turnout.
Total votes cast, Michigan Democratic presidential primary
2004: *
2008: 594,398 (--)
2012: 194,887 (-67%)
2016: 1,205,552 (+519%)
2020: 1,587,679 (+32%)
2024: **768,256 (-52%)
*No state-run primary held
**Preliminary
2016 is an instructive election cycle to examine, because both the Democratic and Republican primaries in Michigan were considered highly competitive. 118,037 more votes were ultimately cast in the Republican primary than the Democratic primary — a total of 9% more primary votes for Republicans than Democrats. This foretold a relative advantage for the Republican nominee, Trump, who went on to win Michigan in the general election. Hillary Clinton garnered 12% fewer general election votes in Michigan than Obama in 2012, while Trump received 8% more votes than Mitt Romney. The 2016 general election performance of both Trump and Hillary basically correlated with trend-lines observable from the 2016 primaries, eight months or so before.
Total votes cast, Michigan Republican presidential primary
2004: *
2008: 869,169 (--)
2012: 996,499 (+15%)
2016: 1,323,589 (+33%)
2020: 683,431 (-48%)
2024: **1,114,151 (+63%)
*No state-run primary held
**Preliminary
Don’t focus so much on individual percentages as the general trend-lines — note how the cycle-to-cycle shifts in primary election performance broadly correlate with candidates’ over or under-performance in the general election. Biden’s ability to win Michigan in the 2020 general election was augured by the 32% increase in total votes cast in the 2020 Michigan Democratic primary. Conversely, the relative disadvantage of Trump was augured by the 48% decrease in Republican primary votes compared to 2016. Biden would go on to receive 535,201 more votes in Michigan in the 2020 general election than Hillary did in 2016 — a 24% increase. Trump also received 370,309 more votes in 2020 than he did in 2016 — a 16% increase, but again dwarfed by the increase in Democratic turnout. (By the way, this cycle-to-cycle increase in Democratic turnout was largely concentrated in affluent suburban areas, not inner-cities — just for anyone out there who might be getting ready to furiously post their comment repeating one of the standard 2020 “election fraud” theories.)
Michigan Democratic primary, 2012
Obama: 89%
Uncommitted: 11%
(Obama +78)
Michigan Democratic primary, 2024
Biden: 81%
Uncommitted: 13%
(Biden +68)
In terms of the “Uncommitted” campaign organized against Biden, the most significant result may not be so much that there was 2 percentage point uptick for “Uncommitted” compared with Obama in 2012, but rather that Biden is exhibiting multi-dimensional under-performance relative to Obama. In aggregate, the Obama 2012 versus Biden 2024 differential translates to Obama having a +10% advantage among Democratic primary voters at this point in 2012. Biden being approximately 10 points less popular than Obama among Democratic voters rings intuitively true to me, and would’ve been close to what I’d have guessed without even seeing any 2024 primary data.
A non-trivial number of Michigan voters used “Uncommitted” this year to register their discontent with Biden, largely for his unswerving support of Israel in the ongoing destruction of Gaza — according to the latest count, 101,436 people voted “Uncommitted.” But in the grand scheme, a +2 point increase for “Uncommitted” between 2012 and 2024 doesn’t tell much of the story of why Biden is currently in trouble. For that, you have to dig slightly deeper into the primary election metrics that have previously had useful predictive power. By most such metrics, Biden is on a trajectory to lose to Trump. That trajectory could always change with future events, foreseeable and unforeseeable, and it was a different story in 2020. But as of today, February 28, 2024, the trajectory is definitely not in Biden’s favor.
Good stuff.
Excellent data. Cheers, Michael.