Harris Campaigned for the Rich, and Everyone Else Will Pay the Consequences

US Nov 5 2024 Election Autopsy - analyzed by {Race} × ({Income} or {Gender})

IMPORTANT! Errata: In figure 1, I had shown Trump gained 3.71m votes from 2020 to 2024 among the white <$50k group. He had actually lost 3.71m votes. This has been updated, and corresponding comments. If you have already loaded this page before, you may need to clear your cookies to see the new figure.

After Harris’s devastating election loss, Pundits have pointed to everything and everyone, from leftists, to Muslims, to "wokeness", to a fundamental racism-misogyny of America. Some of these are completely ungrounded, but some of these have a grain of truth. Yet they miss the broader point. For example, 2024 saw a dramatic decline in black and Latino support from 2020, only white turnout increased (and only among white women did Harris see significant gains); in fact, black women support fell even compared to 2016 - and no-votes "gained" (see Figure 3, Figure 4). The key thing isn’t just how a group voted Dem:GOP this election compared to 2020, but how turnout changed.

Both the Trump and Harris tickets campaigned for the right-leaning middle-class vote - and yes, both made gains (around 4m and 6m respectively), while increasing the group’s turnout. But Harris lost around 10m working class votes (as predicted based on 2020 performance), not to Trump, but to no-shows (see Figure 1) - even under-performing 2016 Clinton (Tables I, II). Notably, white <$50k voters abandoned Harris - but not to Trump (who also lost votes, compared to 2020), but to no-votes (see Figure 1). Harris’s middle-class (and upper-class) gains simply couldn’t offset her incumbency hemmorhaging among the <$50k. If we want to talk about misogynistic-racist gains, this only starts to gain substance for those with $50k+ income; But these MAGA middle class gains (4m) could have been swamped by a working class (11m) brought to ballot box by good economic policy, not Cheney hangouts and Republican talking points. Yet the donors that advised Harris into a diet Republican campaign (and to cut Lina Khan) won’t pay the price; they’ll get the benefits of Trump tax cuts instead (and Lina Khan gone). Fuck you Mark Cuban et al.

Introduction

In the wake of the US election, we have seen many autopsies on what happened - almost all naively taking exit polls at face value, with little consideration of the implications of sectional turnout. As a result, at least online (certainly not the hub of struggling family discourse), battle lines have been drawn - "Latinos getting deported will get what they deserve!". At present, there seems to be a consensus that Trump has "gained" votes, some perhaps defectors from Harris, the latter part of a narrative that Harris has been "abandoned" by 15m votes. In fact, they’ll probably end up around 75m Harris (down 6m from 2020), 79m Trump (up 5m from 2020), with a turnout around 63%. Is this difference a Trump "steal" from the Democrat electorate? Initially I thought so, but now looking at the data, it appears the picture is not so simple.

What happened? Democratic partisans have thrown a slew of answers at the wall: the most absurd is that Arab/Muslim Americans lost them the election - despite their vote contribution not making the difference in a single swing state. So let’s ignore that (though I will briefly address it in the analysis below, to show how unfounded it is - setting aside the absurdity of asking people who’s families, friends, and co-religionists are being genocided with support of the administration Harris is/was VP to). Another is that America is too racist and misogynist. Certainly these factors affected many voters’ decisions. But what about Hillary’s 2016 popular vote win? Obama’s landslide? Does the intersectional discrimnation against black woman explain the Harris catastrophe? Do they think another California slimy liberal politician, if only a white man (ie Gavin Newsom), could run the same diet Republican campaign and win? That such cynical thinking may guide the Democrats is perhaps the real horror story here.

Another related refrain is that Trump has activated the "incel" or "manosphere" or "red pill" community against the, they argue, "DEI hire" of Kamala Harris. So, what is the solution? And who even are these people? Is the idea that we are damned to MAGA election wins because American men are, its argued, flocking towards the Trump-VaPence banner, galvanizing them through their hard-coded chauvinism?

Some have argued that "progressives" are to blame; the bogeyman in their mind: a middle-class white wokester. Yet as we will see (ie look at Fig 1), it appears white voters from $50k+ saw increased turnout for Kamala, compared to 2020 (and among white $50k-$99.999k, even a drop in gross votes for Trump). Now ofc, "progressives" aren’t just among the white middle-class (nor is that result simply due to white middle-class progressive turnout), but it suggests that whole explanation is on shakey ground.

An even more absurd scapegoat: wokeism! Pundits will say that people are sick of "wokeness", invoking some mythical "working class" stereotype that votes on politicians’ virtue signals more than their kitchen table. The argument is rendered more ridiculous by the fact that (A) "wokeness" didn’t yield such catastrophic results in 2020 and 2022, and (B) because the Harris campaign was anything BUT woke, in rhetoric and substance. Anti-migrant, abandoning transgender rights, genocide denial. Is that woke? Then MAGA is woke too!

Bernie and the progressive wing argue that Harris has abandoned the working class. This argument has been treated as dismissive of intersectional concerns over race and class. But, as I hope to make clear here, to talk about class is not to ignore this. Instead, looking at the results by income (above/below $50k income a year) indicates that the election was really only for people making $50k+ a year - if we want to talk about racism and misogny, this is where the substance of it is.

So, what is it?

A fundamental issue is how the exit poll data is analyzed. We see lots of exit poll data showing things like "support for Trump grew in X group". But this misses, a key issue: turnout. And funny enough, we can get an idea of this from those very exit polls. By no means, however, is this a new issue:

In the days after the disastrous 2016 presidential election, a popular meme showing that 94 percent of Black women voters had cast their ballot for Hillary Clinton was circulated as proof that Black women had done their part to keep Trump out of the White House. The meme, though, was misleading. It was true that 94 percent of Black women who voted cast their ballot for Clinton, but those voters represented 64 percent of all eligible Black women. Even though this was a large voter turnout, it represented a 6 percent drop in Black women’s historically high turnout in 2012, when Barack Obama was on the ballot. Indeed, the overall turnout for Black voters declined for the first time in a presidential election in twenty years, falling to 59 percent from its historic high of 66 percent in 2012.

Keeanga-Yamahtta Taylor, Introduction "How We Get Free: Black Feminism and the Combahee River Collective"

Taylor’s point here is ever-present in US elections - we shouldn’t take the ratios in election exit polls simply at face value. Perhaps even more significant is the turnout. For example, the US eligible voter population was 13.5% and 14% black in 2020 and 2024 respectively. In 2020, they were about 13% of those in the CNN exit poll - representing a turnout of about (13/13.5)*66.6% = 64.1%). But in 2024, black voters were only 11% of those in the CNN exit poll, indicating a significant drop in black turnout: (11/14.0)*0.63 = 49.5%. Taking the black eligible voter population of 2022, 34.45m, that’s a drop of (0.641-0.495)*34.45 = 5.0m black voters, from 2020 levels. Following Taylor above, this would be the second drop in black voter turnout, since 2016, which was then the "first time in twenty years". In general, Latino, Asian, and "Other racial/ethnic groups" turnout also fell, and quite dramatically. The only racial group to see increased turnout in 2024 were whites.

In this article, I look at how, based on exit poll data from CNN between combinations of {black/white/Latino}, {male/female}, and income groups, the vote changed from 2020 to 2024. Some of this analysis may ultimately need correction, as exit polls are imperfect, and more precise values of ultimate turnout are made apparent (my prediction was around 63% turnout). Nonetheless, they do have a substantial sample size, and I compare results from 2016, 2020, and 2024 CNN exit polls, to account for any systematic errors that may be present. Therefore, the results may not be 100% accurate as absolute values, but likely the trends are indicative. (See "Data, Terminology, and Methods" below for more information)

While 2020 was an exceptional year, in 2024, we have still seen high turnout for this century. Despite this, the Democrats were dealt a resounding defeat. Why? My exit poll analysis shows this was an election for three groups:

1. The rich - Across income groups, only the wealthiest Latinos ($100k+ turnout shift: 34% → 49%), the wealthiest whites ($100k+ turnout shift: 44.9% → 67.2%), and the wealthiest blacks (25.2%(?) → 51.4%) did turnout increase (also, white $50k-$99.999k stayed around 80% turnout, the only race-income group to stay at such a high level). Except black people making <$50k (which stayed roughly the same as 2020), all other racial-income groups saw a gross increase in no-voters (see Figure 1). In general, for 2016, 2020, 2024, the turnout for <$50k was 60.8%, 71.9%, and 54.3%; for $50k+ the turnout was 59.4%, 63%, and 66.7%.
2. White women - across nearly all racial-gender groups, all saw a gross increase in non-voters - white male: +2.22m, black male: +0.9m, black female: +4.99m, latino male: 0.77m, latino female: +5.15m. For black and Latino voters, female turnout fell significantly more than male turnout (about 5x and 6.4x more no-votes for blacks and Latinos respectively, even after accounting for gender differential eligibity (more women are generally eligible than men)). All except white women, among which there were 6.37m fewer non-voters than in 2020, and the only race-gender group which Harris saw an increase in votes over 2020 (see Figure 3). In fact, even compared to 2016, many of these trends hold (black women’s support of the Democratic candidate falls from 2016 to 2020, even a marginal increase of support for Trump!) (see Figure 4).
3. White people - This is basically implied by the above point, but it is worth mentioning. Based on exit poll data, black turnout fell 14.6% (64.1% → 49.5%), and Latino turnout fell 12.3% (63.7% → 51.4%) (these are even lower than 2016, when black turnout was 53.4%, and Latino 55.6%). Only white turnout increased, by 2.5% (66.1% → 68.6%; compare this with 2016, at 61.1%).

All of this raises some serious issues with the preferred liberal arguments as to why Harris lost (ie misogyny and racism, that she was somehow "too left wing", that [middle class] progressives are to blame, that whites were offput by her being a black woman, etc). For one, the POC female vote fell harder for her than the POC male vote. POC males did vote more Trump than before (we will see the Trump increase among them is not simply a mirage of lower Dem turnout), but a significant factor in Harris’s defeat is the net loss of nearly 3m POC women voters from 2020. Additionally, the biggest turnout shortages, in terms of race, were amongst Latinos and black people.

If the prevailing explanation for Harris’s loss is racism and misogyny (a factor certainly, but proponents here argue its the factor), why then did black women’s support of Harris fall, not only compared to 2020 (a catastrophic decline compared to then, a drop of 4.8m votes (at least, based on the exit poll data analysis)), but also, it appears, compared to 2016 (Figure 4)? A year in which, as Taylor points out, was the first time in twenty years that black turnout had fallen. Should we argue that, rather than black women being demoralized by her horrible campaign, that this shows many black women are "self hating"? (since, as we all have been told since the election, everyone is at fault but Harris!) Had Harris had those 4.8m lost votes from black women (had things went as well as 2020), she would have at least won the popular vote. Perhaps this drop is partially due to GOP policies to limit black voter registration, but the drop is huge - I wouldn’t buy the idea that this is purely a result of such policy, as opposed to a drop in voter engagement. This is not to disparage black women. Instead, this observation should (A) be put under scrutiny as better data rolls in, and (B) highlights how the identity view of politics has deeply damaged the Democratic Party’s capacity to win elections. Not simply because many voters don’t like it, but because it fundamentally misguides the Party on how to win elections, even among the identity groups they allege to promote and support. Ironically, it was white women who were most energized by the Harris campaign (though their overall voter ratio still favors Trump, higher Harris support among white women appears to be due to increased voter turnout; both Democrats and Trump have seen increasing support from white women in 2020 and 2024).

And the elephant in the room: the working class. For those making less, neither candidate saw gains in gross votes from 2020 - instead, turnout declined for both parties, only it declined faster for the Democrats. This shouldn’t be a surprise - Harris made no economic arguments on the campaign, while she is the incumbent vice president in an economy that many <$50k feel hardship under - and that the Biden administration, and liberal wonks on TV, have constantly said isn’t actually bad. As a group, Harris lost 11m votes here, but that didn’t translate to a Trump gain (he lost around 0-4m votes): just the number of people in this group that simply didn’t vote grew a lot. Turnout declined from 71.9% in 2020 (when Biden was making an economic populist campaign, amidst COVID fears, of unprecedented degree since perhaps LBJ in the 1960s) to 54.3%. As a result, it appears Trump won this group - but he didn’t gain votes; he lost them slower than Harris.

Her losses were particularly acute with the white working class. She didn’t lose voters to Trump though - he also lost voters. Both lost voters to no-vote. Similar losses were registered in the Latino sub-$50k group, although with a slight Trump gain instead (but far outweighed by the gain in no-vote). Notably, the black sub-$50k group stayed largely constant, with a slight drop in Trump support and a slight increase in Harris support. Overall, only among $100k+ Latinos, $50k+ whites, and <$50k blacks did Harris see any gross vote gains (and perhaps among $100k+ blacks as well): and only among $100k+ whites (and maybe among $100k+ blacks) was this substantial.

By contrast, both candidates gained votes among the $50k+ crowd, and Harris outpaced Trump (this was the basis of her whole Mark Cuban campaign strategy). But she didn’t gain enough to compensate for the collapse in <$50k voter participation. But considering this is where Trump made substantive gains (and keep in mind, the Jan 6th riot in the Capitol was largely a MAGA middle class affair), if we want to talk about misogyny and racism driving a MAGA vote, this is where to look - America’s upper class, and yes, its middle class (this $50k+ group also includes some working class ofc, income isn’t a perfect discriminator; I use it because it’s the threshold used by many exit pollsters). That hallowed middle class that Harris tried so hard to win - and she kinda did. All of this reinforces the debates spawned since Riley and Brenner’s "Seven Theses on American Politics" in NLR, which argued that American politics is a conflict between two coalitions of the American bourgeoisie.

But beneath this is the fact that the US working class is a crucial electorate, despite relatively low voter turnout, and traditionally one that favors the Democratic Party. By leaving the working class behind - 11m votes had she performed similar to 2020 - Harris lost the election spectacularly, crippling the Democratic party downballot across the nation, and delivering MAGA fascism a House and Senate majority, along with an executive office and SCOTUS.

Now, the donors who begged her to avert from an economic populist campaign (as signalled by her Tim Walz VP pick) will go back to enjoy their tax cuts, and we will suffer the consequences. She outspent Trump by nearly 3x (about $400m to $1.2b), and look what we got (although maybe Trump’s team outspent Harris’s by $40bn - the price of Elon Musk turning Twitter into a reactionary hellhole).

For this article, I followed some predictions by Norm Finkelstein, when he predicted that Harris would lose bc (A) she has no message, (B) the white working class would abandon her, (C) the youth vote would simply not vote, and (D) the Latino vote could go either way. We will find these were remarkably astute predictions, along with the other significant results.

Before going into the data analysis, a few remarks. (1) is the need for a "leftist" economic program, perhaps "populist". Whatever you want to call it. While progressives are one of the maligned groups by angry Harris supporters, the problem wasn’t simply an ideologically bitter "left-wing" of the Party, but a working class deeply frustrated with their current situation. Pundits will say the politics of Bernie would be unpopular, but considering that Biden incorporated some of Bernie’s politics in his 2020 campaign, and won huge turnout among the <$50k bracket (see Table II), perhaps that should be reconsidered. While as a committed socialist I would agree that Bernie’s politics are more "left-wing" than the current status quo, this left-right dichotomy is, for most people, irrelevant. That’s why, for example, so many seem to like both Trump and Bernie (and AOC). Not because they identify with "leftism" - nor because "leftism" "horseshoes" with far-right politics. But because people want to see a change from the current system, they’re mad at it, and either don’t vote, or end up voting for "change" (Trump). While putting politics on a spectrum might have some analytic utility, it apears nearly useless in predicting how people engage with it. At least, a people who have been deliberately rendered politically ignorant by both parties ("left wing when DEI, right wing when no DEI"), and a working class who have been rendered increasingly unorganized since the 1970s (ie, declining union membership and strength).

More insidiously, (2), does the Democratic Party know identity politics campaigns would perform worse in elections, compared to economic populist campaigns? And they opt for the "losing strategy" because (A) their donors don’t want economic populism and (B) because, when they throw the electoral dice, they prefer to lose than see actual change that would help working Americans, even if it meant the most horrible thing in the world - taxing the rich? It’s worth recalling that the Democratic Party deliberately organized to stop a Bernie primary win in 2020. Is it really simply ignorance? Either way, this election, in my view, reveals the deep failure of the Democratic Party’s guiding electoral tactics.

Finally, (3), this was simply the most horrible Democratic election campaign in my life. Harris’s denial of the Gaza Genocide, and active campaign-trail support of Israel’s conduct, was atrocious. While the main defense for her not "breaking" with Biden was out of decorum for her VP office (a disgusting defense for continued genocide), she did say she was willing to break with Biden on some things: for example, on The View, she said she would be different from Biden by appointing a Republican to her cabinet. She was willing to break with Biden, not in a humanitarian direction, but to become even more right-wing. Even aside from her insulting campaigning on the basis of Bush and Cheney support, her win would likely not bring an end to the genocide. As Adam Johnson, of the Citations Needed podcast, has observed (The Nation, The Real News Network), the Biden administration’s idea of "ceasefire" was a mockery, devised only to confuse voters into thinking it was something substantial. That this was the attitude has basically been revealed in the past few months. Further, Harris’s campaign supported leaving transgender rights up to states (the same slimy argument Trump made for abortion; abortion rights being, apparently, the only popular issue Harris was willing to campaign on (and an issue which she was far outpaced by on ballot initiatives across the country, ie FL: support of abortion was 57.2%, for Harris, 43%; 2020 Biden: 47.86%)), and supported a far-right policy on the border and immigrants. Her outreach to black men? Protecting their cryptocurrency investments. For the working class? Nothing. For women of color? Her campaign for them was more-or-less solely the fact that she would be the first woman of color US president (although it’s worth remarking that she made virtually no gestures towards the identity politics aspect of her campaign; it was supposed to simply speak for itself). The only material promises were made to the middle - and especially upper - class. This far-right, genocide-denying-and-enabling campaign was doomed from the start. "Wokeism" was the problem, my ass.

First, the Muslim/Arab/Third Party Vote

(Back to Table of Contents)

Along with Latinos, chauvinistic Harris supporters have also unleashed disgusting bile on Arab/Muslim voters (and the Third Parties they voted for). In fact, if we look at key swing states, we will find no example of a third party spoiler (following data from AP, states at 99%, except AZ, at 98% counting):

• MI: Trump - 2804647 votes; Harris - 2724029 votes; Δ = 80618; Green Party Vote - 44697
• PA: Trump - 3536236 votes; Harris - 3409373 votes; Δ = 126863; Green Party vote - 34364
• WI: Trump - 1697298 votes; Harris - 1667881 votes; Δ = 29417; Green Party votes - 12266
• NC: Trump - 2897782 votes; Harris - 2714346 votes; Δ = 183436; Green Party votes - 24752
• GA: Trump - 2663110 votes; Harris - 2548014 votes; Δ = 115096; Green Party votes - 18229
• AZ: Trump - 1756999 votes; Harris - 1571066 votes; Δ = 185933; Green Party votes - 18122
• NV: Trump - 750619 votes; Harris - 704985 votes; Δ = 45634; Green Party votes - not on ballot?

Unsatisfied with this, nitpickers (ie people who militantly refuse to criticize the Harris campaign at all for losing the worst for Democrats since 2004, and delivering MAGA down-ballot wins across the country, such that they secure the House and Senate) have observed that total Arab/Muslim votes in Michigan for Trump and third parties surpass the difference there. For example, this Conversation article, the author analyzes voters in Dearborn, Dearborn Heights, and Hamtramck, observing Trump won 50332 votes. Notice with the Green Party votes, that seems to make the difference (50332+44697 = 95029 > 80618). What this misses (assuming that the Green Party vote total is largely from Muslim/Arabs, which I will for the sake of the argument) is that some in these areas were already Trump voters. If the argument is "angered by the Gaza Genocide, Arab/Muslim voters switched from the Democratic Party", then the already-Trump voters have to be left out of consideration, since they probably weren’t going to vote Democratic, regardless of anything in Gaza (except maybe switching to Democratic, had the Biden administration taken the humane and anti-genocidal course there).

For example, if we look at 2020 results in Dearborn MI, we see Biden won 30.7k votes, and Trump 13.2k votes. In 2024, Harris won 15k votes here, and Trump 18k votes. Thus, Trump "only" gained 4.8k votes (ie 27% of his votes were "new"). Since Dearborn’s 18k Trump votes are about 36% of those analyzed by the Conversation article, let’s take those as indicative of the overall pattern. That is, that only 27% of those 50332 votes were new, or 13590 new Trump votes. With the Green Party’s statewide 44697 votes (again, leaving aside if these are all Muslim/Arab, and ignoring that there were already Green voters before, and ignoring that some of those voters may have never voted before for anyone, given the US’s relatively low turnout rate), that’s at most 13590+44697=58287 votes of Arab/Muslim Michiganders who switched from Democratic to Trump, insufficient to overcome the 67028 threshold (since if Trump lost those 13590 voters, the difference would go down to 67028 votes).

So no, enraged Arab/Muslim voters didn’t lose Harris the election, unless we’re counting voters in Dearborn+Dearborn Heights+Hamtramck who were already voting Trump. And even if they did, that wouldn’t have won Harris the election, nor reversed her huge popular vote loss!

Finally, this isn’t to say Arab/Muslim voters should have voted Harris. She is/was vice president of an administration arming and supporting Israel’s genocide in Gaza. To boot, she continued to deny the genocide throughout the campaign, and blocked out Arab/Muslim speakers from events such as the Democratic National Convention. It is absurd to have expected them to vote "blue no matter who".

With that aside, on to the "real" analysis.

Working Class as a Whole

(Back to Table of Contents)

Finkelstein: will lose the white working class vote

This CNN exit poll indicates whites making <$50k (25% of respondants), went 63 Trump:35 Harris ($50k+: 51 Trump, 46 Harris). This is a dizzying result, but I want to look at the US "<$50k" group more broadly. I think this indicates a fundamental lesson from the campaign. Later on, we will return to the whites based on income brackets. But we will then see that both Harris and Harris lost gross votes from this group (see Figure 1, Section: White). Trump’s lead here is a residue of (A) prior support and (B) declining turnout overall (ie Trump’s share is big amongst a shrinking group of white <$50k voters; this especially so, because Trump lost voters here slower than did Harris).

Across all groups, <$50k went Trump 50% to Harris 47% (27% of voters); for $50k+: 48% Trump, 49% Harris. In 2020, <$50k+ went 55 Biden, 44 Trump; $50k+ went 48 Biden, 49.2 Trump. Thus we see below $50k, setting aside turnout questions, Harris saw a drop of 8 points, and marginal gains in the $50k+ group. It also shows that 35% of respondents were <$50k, and 65% were $50k+. In the 2024 exit poll, 27% were <$50k, and 73% were above. At the same time, in 2020, according to the US Census Office, 31.3% made <$50k (and thus 68.7% were over) in 2023. In 2020, the numbers are 32.4% (and thus 67.6% were over). That means 2024 <$50k turnout was 54.3%; for over $50k, it was 66.7%. In 2020, <$50k turnout was 71.9%, and over $50k turnout was 63%.

Here are the results for 2016:

Table I - 2016 US Presidential Election, income group results; turnout 60.1%; voting-eligible population was 230,931,921 people; population $50k: 64.5% (148.95m); turnout for $50k+ 59.40%; total votes for $50k+: 41.32m:42.57m:4.91m:60.13m

Item	<$50k	$50k-$99.999k	$100k+
Population	82.21m	64.89m	84.06m
Population Percent (PP)	35.6	28.1	36.4
𝑛 % (RP)	36	30	34
𝑛 - 24558	8841	7367	8350
Turnout - (RP/EP)*60.1	60.8	64.2	56.1
Dem: GOP: Other	53: 41: 6	46: 49: 5	47: 47: 6
Dem: GOP: Other: None	32.2: 24.9: 3.6: 39.2	29.5: 31.5: 3.2: 35.8	26.4: 26.4: 3.4: 43.9
Dem: GOP: Other: None total	26.49m: 20.49m: 3.00m: 32.23m	19.16m: 20.41m: 2.08m: 23.23m	22.16m: 22.16m: 2.83m: 36.90m
Total (w/o 3rd Party)*	79.21m	62.8m	81.22m

*Third Party is excluded here to be consistent with other tables.

And here are the results for 2020 and 2024:

Table II - 2024 US Presidential Election, vote breakdown by income; results reported as 2024-data (2020-data); 2020 turnout: 0.666; 2024 turnout: 0.63; eligible populations: 2020: 238.3m; 2024: 246.05m; $50k: population (from prior analysis, needs update with more preicse results below): 167.63m (164.94m); Percent Surveyed: 73(65); Turnout: 66.7 (63.0); Dem:GOP: 49:48 (48: 49.2); Dem:GOP:None: 32.7: 32.0: 33.3 (30.2: 31.0: 37); Dem: GOP: None total: 19.5m: 20.7m: 34.9m (30.5m: 24.4m: 21.7m); Total: 164.3m (157.9m); Dem: GOP: None total Δ(24, 20): +6.1m: +4.0m: -3.7m; Dem: GOP: None total Δ (24,16): +13.48m: +11.13m: -4.33m; Dem: GOP: None ratio Δ(24, 16): +5.0: +3.4: -7.1

Item	<$50k	$50k-$99.999k	$100k+
Percent Population (PP)	31.3 (32.4)	27.8 (27.5)	40.9 (40.0)
Population	76.37m (77.11m)	68.40m (65.53m)	100.63m (95.32m)
Percent Surveyed (PS)	27 (35)	32 (39)	40 (26)
Turnout (turnout*(PS/PP))	54.3 (71.9)	72.5 (94.5)	61.6 (43.3)
Dem: GOP of voters (DG)	47: 50 (55: 44)	46: 51 (57: 42)	51:46 (42: 54)
Dem: GOP: None of all (DGN)	25.5: 27.2: 45.7 (39.5: 31.6: 28.1)	33.4: 37.0: 27.5 (53.9: 39.7: 5.5)	31.4: 28.3: 38.4 (18.2: 23.4: 56.7)
Dem: GOP: None total of all (tDGN)	19.5m: 20.7m: 34.9m (30.5m: 24.4m: 21.7m)	22.81m: 25.29m: 18.81m (35.30m: 26.01m: 3.60m)	31.61m: 28.51m: 38.64m (17.33m: 22.29m: 54.05m)
Total (w/o 3rd Party)	75.1m (76.6m)	66.91m (64.91m)	98.76m (93.67m)
Dem: GOP: None total Δ (24,20)	-11m: -3.70m: +13.2m:	-12.49m: -0.72m: +15.21m	+14.28m: +6.22m: -15.41m
Dem: GOP: None total Δ (24,16)	-6.99m: +0.21m: +2.67m	+3.65m: +4.88m: +11.67m	+9.45m: +6.35m: +1.74m
Dem: GOP: None ratio Δ (24,16)	-6.8: +2.3: +6.5	+3.9: +5.5: -8.3	+5.0: +1.9: -5.5

Looking at Table II, we can see that (0.719-0.543)*0.313 = 5.5% of the whole US population didn’t vote in 2024, than (roughly) did in 2020, simply from the <$50k group. From $50k-$99.999k, (0.945-0.725)*0.278=6.1% - altogether, 11.6%. In fact, this is where the bulk of the turnout dropoff happens; among $50k+ group, we see a population-wide increase of (0.667-0.630)*0.687 = 2.5%. This indicates a total drop of 3% from 2020, or 66.6%-3%=63.6%, within our expectations of changed voter turnout. Since the eligible electorate is around 244m in 2024, that’s around 13.4m fewer <$50k than 2020 group turnout would predict, and 6.2m more $50k+ than 2020 group turnout would predict (and specifically, the action is in $100k+).

Looking at the estimated totals for the country for <$50k, we see Trump loses 3.7m votes from 2020; the monkey wrench is Harris lost even more votes, 11m - and no-vote gained 13.2m (Note that this group was 77.1m eligible voters in 2020, and 76.4m in 2024; given their 2020 performance, and at 2020 turnout for this group, we would expect this slight decline would mean Trump loses 0.23m votes here, and Harris loses 0.29m votes). Even compared to the 2016 election, Democrats lost 7m votes. While the ratio of Dem support here fell by 6.8 percent points (pp), the no-vote pp increased 6.5. Had the 2024 results here been even that 6.5 pp higher (so, still not even as popular with <$50k as Hillary Clinton), they would have had 0.065*76.37 = 4.96m more votes. Even this number, far smaller than the fall from 2020, is still more than Trump’s gain with $50k+ from 2020 - and barely lower than where Trump performed best in 2024, among $100k+. Further, looking at 2016, it appears plausible Trump hasn’t really made headway with <$50k since 2016, with roughly the same total votes in this group in 2016 and 2020.

At the same time, Trump gained 6m voters among the $100k+ group, but Harris gained 14.28m - and no voters fell by 15.41m (notice these are different from the 2020 group turnout predictions, because the size of these groups has changed; the above predictions are to say "what if X grouped behaved the same in 2024 as 2020"). Altogether, Harris loses around 5m, and Trump gains 1.8m votes; the results for Harris are consistent with results, although for Trump, there’s a missing ~2.2m - perhaps he was under-polled in one or both groups.

Even if all of that "missing 2.2m" was in <$50k, that would put him at 22.9m there - still falling short. What this indicates is not so much a Trump landslide, but two parallel elections: in election 1 (for <$50k, and among $50k-$99.999k), Trump’s performance declined. But Harris’s performance plummeted, losing 11m votes, 36% of the 2020 total, in <$50k alone. Naturally, that means voter turnout plummeted. Election 1 wasn’t a "real" election - this was the group left behind; neither candidate was campaigning here, while they simultaneously felt their economic hardship was the responsibility of the Biden Administration. Justified or not, the Harris campaign made no effort to argue against this, or to provide an economic plan that showed a way out. As a population (although ofc, many voters probably switched to Trump, and as many or more away from Trump, so not nobody switched to him), the <$50k and $50k-$99.999k group didn’t "move" to Trump - they left Trump slower than Harris. Perhaps this is inflected with racism and misogny - but the fact that (A) they didn’t move to Trump (except as a proportion in a shrunk turnout) and (B) that this group, by definition, will have "bread-and-butter" issues more on the mind indicates this is a broader failure.

In election 2, both candidates saw gains. This means these were the votes the two candidates were competing for. If we want to talk about "people drawn to vote for Trump because of X reason" (ie misogny, racism, etc), this is the group - and a group Harris could have overcome by mobilizing the <$50k voters. Now this isn’t to say $100k+ behaves the same, but it does indicate we need to differentiate our critique of the racist-misogynist voter in terms of income, and more bluntly, class.

The <$50k and $50k-$99.999k group was simply left behind - 11m votes in <$50k alone - and 23.5m altogether - that Harris could, if 2020 is any indication, have won. Had she gained that 11m, her total vote would have increased by 15%; that 23.5m, 31.3%! In swing states around the country, margins were tight: 131k in PA (19), 80k in MI (15 EC), 30k in WI (10), 184k in AZ (11), 115k in GA (16), 190k in NC (16) - a total "swing" differential of 730k votes, or just 6.6% of that "forgotten" 11m, 3.1% of the "forgotten" 23.5m. This is a gigantic lost opportunity - the cost of campaigning as a Republican, with no economic platform.

Racial Groups by Gender and Income Bracket

In this analysis, I will look at how race-income and race-gender groups voted in 2016, 2020, and 2024, for white, black, and Latino. However, the 2016 CNN exit polls don’t show results for race-income (whereas their exit polls for 2020 and 2024 do). So I will first present the overall race and race-gender results for 2016 in Tables III and IV, before going into the white, black, latino -income and -gender results for 2020 and 2024.

2016 Race and Race-Gender Results

(Back to Table of Contents)

Table III - 2016 US Presidential Election, race results; turnout 60.1%; voting-eligible population was 230,931,921 people

Item	White	Black	Latino	Asian
Eligible Population	161.3m*	31.0m	27.3m	11.3m
Eligible Percent (EP)	69.8*	13.5	11.9	4.9
𝑛 % (RP)	71	12	11	4
𝑛 - 24558	17436	2947	2701	982
Turnout - (RP/EP)*60.1	61.1	53.4	55.6	49.1
Dem: GOP: Other	37: 57: 6	89: 8: 3	66: 28: 6	65: 27: 8
Dem: GOP: Other: None	22.6: 34.8: 3.7: 38.9	47.5: 4.3: 1.6: 46.6	36.7: 15.6: 3.3: 44.4	31.9: 13.3: 3.9: 50.9
Dem: GOP: Other: None total	36.47m: 56.18m: 5.91m: 62.75m	14.73m: 1.32m: 0.50m: 14.45m	10.02m: 4.25m: 0.91m: 12.12m	3.61m: 1.50m: 0.44m: 5.75m
Total (w/o 3rd Party)*	155.4m	30.5m	26.39m	10.86m

*Third Party is excluded here to be consistent with other tables.

The following is the race-gender results for 2016. I split the table into IV.1 and IV.2 for better viewability on a mobile device.

Table IV.1 - 2016 US Presidential Election, race-gender results (White/Black Men/Women); turnout 60.1%; voting-eligible population was 230,931,921 people

Item	White Men	White Women	Black Men	Black Women
Eligible Voter % (gEV*EP)	34.2%	36.2%	6.3%	7.2%
𝑛 % (RP)	34	37	5	7
𝑛 - 24558	8350	9086	1228	1719
Turnout - (RP/EP)*{Tᵢ}	59.7%	61.4%	47.7%	58.4%
Dem: GOP: Other	31: 62: 7	43: 52: 5	82: 13: 5	94: 4: 2
Dem: GOP: Other: None	18.5: 37.0: 4.2: 40.3	26.4: 31.9: 3.1: 38.6	39.1: 6.2: 2.4: 52.3	54.9: 2.3: 1.2: 41.6
Dem: GOP: Other: None total	14.62m: 29.23m: 3.30m: 31.83m	22.07m: 26.69m: 2.57m: 32.27m	5.69m: 0.90m: 0.35m: 7.61m	9.13m: 0.39m: 0.19m: 6.92m
Total (w/o 3rd Party)*	75.68m	81.03m	14.2m	16.44m

*Third Party is excluded here to be consistent with other tables.

Table IV.2 - 2016 US Presidential Election, race-gender results (Latino Men and Women); turnout 60.1%; voting-eligible population was 230,931,921 people

Item	Latino Men	Latino Women
Eligible Voter % (gEV*EP)	5.8%	6.1%
𝑛 % (RP)	5	6
𝑛 - 24558	1228	1473
Turnout - (RP/EP)*60.1	51.8%	50.1%
Dem: GOP: Other	63: 32: 5	69: 25: 6
Dem: GOP: Other: None	32.6: 16.6: 2.6: 48.2	34.6: 12.5: 3.0: 49.9
Dem: GOP: Other: None total	4.37m: 2.22m: 0.35m: 6.46m	4.87m: 1.76m: 0.42m: 7.03m
Total (w/o 3rd Party)*	13.05m	13.66m

*Third Party is excluded here to be consistent with other tables.

Latino

(Back to Table of Contents)

Finkelstein: Ambiguous on Latino voters; sees lots of dissatisfaction

2020 Latino turnout: (13/13.6)*66.6 = 63.7%; 2024: (12/14.7)*63 = 51.4%; pew eligible Latino voters

Table V - 2024 US Presidential Election, Latino vote breakdown by gender; results reported as 2024-data (2020-data); 2020 turnout: 0.666; 2024 turnout: 0.63; gEV: male:female = 49:51

Item	Male	Female
Eligible Population ({36.2m/32.3m}*{0.49/0.51})	17.7m (15.8m) (Δ=1.9m)	18.5m (16.3m) (Δ=2.2m)
𝑛 - 2152 (2451)	1011 (907)	1141 (1544)
Percent Eligible Voters (PEV) (14.7/13.6)	5.8 (3.9)	6.2 (6.3)
Percent Surveyed (PS) (s)	47 (37)	53 (63)
Turnout ({0.514/0.637}*(PS/gEV))	49.3% (48.1%)	53.4% (78.7%)
Dem: GOP of voters (DG)	43: 55 (59: 36)	60: 38 (69: 30)
Dem: GOP: None of all (DGN)	21.2: 27.1: 50.7 (28.4: 17.3: 51.9)	32.0: 20.3: 46.6 (54.3: 23.6: 21.3)
Dem: GOP: None total of all (tDGN)	3.8m: 4.8m: 9.0m (4.5m: 2.7m: 8.2m)	5.9m: 3.8m: 8.6m (8.9m: 3.8m: 3.5m)
Total (w/o 3rd Party)	17.6m (15.40m)	18.30m (16.20m)
Dem: GOP: None total Δ (24, 20)	-0.73m: +2.06m: +0.77m	-2.92m: -0.09m: +5.15m
Dem: GOP: None ratio Δ (24, 20)	-7.2: +9.8: -1.2	-22.3: -3.3: +25.3
2016: Dem: GOP: None total	4.37m: 2.22m: 6.46m	4.87m: 1.76m: 7.03m
Dem: GOP: None ratio Δ (24, 16)	-11.4: +10.5: +2.5	-2.6: +7.8: -3.3

Some observations: Yes, Latino men did see a gain of 2.1m votes for Trump from 2020, a 78% increase from that year. At the same time, Harris lost 0.7m votes among them, and the pool of no-vote grew by 0.8m. Among Latina women, Harris’s support fell drastically by 3m, while Trump stayed constant at about 3.8m. The pool of no-vote grew by 5.6m.

Looking at 2016, Trump has gained around 2.6m male Latino voters, and the Dems have lost 0.57m (no votes increased 2.54m), with an overall slight turnout decrease. Among Latino women, the Dems have gained about 1.03m voters, Trump has gained 2.04m voters, and no-votes gained 1.57m voters - with an increase in turnout since 2016.

Further, let’s look at Latino by income:

Table VI - 2024 US Presidential Election, Latino vote breakdown by income; results reported as 2024-data (2020-data); 2020 turnout: 0.666; 2024 turnout: 0.63;

Item	< $50k	$50k to $99.999k	$100k+
𝑛 - 2152 (2451)	624 (735)	818 (1201)	710 (515)
Percent Surveyed	29% (30%)	38% (49%)	33% (21%)
Percent of Hispanic Population (PHP)	34.7% (32.3%)	30.8% (28.2%)	34.6% (39.4%)
Turnout (PS/PHP)*{63.7/51.4}	43.0 (59.2)	63.4 (96.6)	49.0 (34.0)
Dem: GOP of voters (DG)	43: 55 (64: 33)	55: 45 (74: 26)	42: 56 (38: 61)
Dem: GOP: no-vote (DG)	18.5: 23.7: 57.0 (37.9: 19.5: 40.8)	34.9: 28.5: 36.6 (71.5: 25.1: 3.4)	20.6: 27.4: 49.0 (12.9: 20.7: 66.0)
Dem: GOP: no-vote totals (DG)	2.32m: 2.97m: 7.16m (3.95m: 2.04m: 4.26m)	3.89m: 3.18m: 4.08m (6.51m: 2.29m: 0.31m)	2.58m: 3.44m: 6.39m (1.64m: 2.64m: 8.40m)
Total (w/o 3rd Party)	12.45m (10.25m)	11.15m (9.11m)	12.41m (12.68m)
Dem: GOP: None totals Δ (24, 20)	-1.63m: +0.93m: +2.90m	-2.62m: +0.89m: +3.77m	+0.93m: 0.80m: -2.01m
Dem: GOP: None ratio Δ (24, 20)	-19.4: +4.2: +16.2	-36.6: +3.4: +33.2	+7.7: +6.7: -15.0

Asian

(Back to Table of Contents)

Note: I would like to look at the Asian demographic (I ought to, I am one :P), but the Asian vote is a relatively small chunk compared to Latino, White, Black. For example, in 2020, we were 5.5% of eligible voters, and 6.1% in 2024. Based on the CNN exit polls, this translates to a 2020 turnout of (4/5.5)*66.6 = 48.4%. For 2024, (3/6.1)*63 = 31.0%.

More fundamentally is the problem of small numbers in the exit polls. However, just 3% of for the 2024 exit poll (thus 22914*0.03=688 participants), and 4% of 2020 exit poll (15590*0.04 = 624 participants) means the results are already subject to error from small numbers - it’s possible Asian turnout was quite different from 48.4% and 31.0% as a result (ideally, a population-level sample should have at least ~1000 people). And this issue will be even worse when looking at Asians broken down by gender and income groups.

White

(Back to Table of Contents)

Let’s return here to Finkelstein’s prediction, that the white working class will abandon Harris.

Regarding data, I’ve, funny enough, had a hard time finding how many eligible voters are white - but can fnd how many are Black, Asian and Hispanic (BAH) (which are some of the largest minorities) (ie here). Altogether, there were 246.05m eligible voters in 2024, and 238.3m eligible voters in 2020. Of these, 85.65m were BAH in 2024, and 77.45m in 2020. Assuming white’s form a large bulk of the remainder, we will correlate white’s with the difference: 160.85m in 2020 (67.5% "white") and 160.4m in 2024 (65.2% "white"). The general gender eligible voter ratio is 49M:51F, so I will use that here. If you have direct data for the white eligible voter population for 2016, 2020, and/or 2024, please email me at glaznaruost@gmail.com.

In 2020, 67% of respondants were white, so white turnout was (67/67.5)*0.666 = 66.1%; 2024, 71% of voters were white, so turnout was (71/65.2)*0.63=68.6%.

Table VII - 2024 US Presidential Election, White vote breakdown by gender; results reported as 2024-data (2020-data); 2020 turnout: 0.666; 2024 turnout: 0.63; geV: male:female = 49:51

Item	Male	Female
𝑛 - 16848 (9920)	8087 (5158)	8761 (4762)
Eligible Population ({160.4/160.85m}*{0.49/0.51})	78.6m (78.8m)	81.8m (82.0m)
Percent Surveyed (PS) (s)	48 (52)	52 (48)
Turnout ({0.686/0.661}*(PS/gEV))	67.2% (70.1%)	69.9% (62.2%)
Dem: GOP of voters (DG)	37: 60 (38: 61)	45: 53 (44: 55)
Dem: GOP: None of all (DGN)	24.9: 40.3: 32.8 (26.6: 42.8: 29.9)	31.5: 37.0: 30.1 (27.4: 34.2: 37.8)
Dem: GOP: None total of all (tDGN)	19.54m: 31.69m: 25.78m (20.99m: 33.70m: 23.56m)	25.73m: 30.30m: 24.62m (22.44m: 28.05m: 31.00m)
Total (w/o 3rd Party)	77.01m (78.25m)	80.65m (81.49m)
Dem: GOP: None total Δ (24, 20)	-1.45m: -2.00m: +2.22m	+3.29m: +2.25m: -6.37m
Dem: GOP: None ratio Δ (24, 20)	-1.7: -2.5: +2.9	+4.1: +2.8: -7.7
2016: Dem: GOP: None total	14.62m: 29.23m: 31.83m	22.07m: 26.69m: 32.27m
Dem: GOP: None ratio Δ (24, 16)	+6.4: +3.3: -7.5	+5.1: +5.1: -8.5

Comparing with 2016 results (Table IV), we see white males gave Dems 4.92m more votes since 2016, 2.46m more votes to Trump, and 6.05m less no votes, with an increased turnout. The overall ratio has increased slightly in favor of Trump. For white women, Dems gained 3.66m votes, Trump gained 3.61m votes, and no votes lost 7.65m votes, with turnout increasing.

Table VIII - 2024 US Presidential Election, White vote breakdown by income; results reported as 2024-data (2020-data); 2020 turnout: 0.666; 2024 turnout: 0.63;

Item	< $50k	$50k to $99.999k	$100k+
𝑛 - 16848 (9920)	4212 (3571)	5391 (3373)	7413 (2976)
Percent Surveyed	25% (36%)	32% (34%)	44% (30%)
Percent of White Population (PWP)	27.8% (28.9%)	27.2% (26.9%)	44.9% (44.2%)
Turnout (PS/PHP)*{68.6%/66.1%}	61.7% (82.3%)	80.7% (83.5%)	67.2% (44.9%)
Dem: GOP of voters (DG)	35: 63 (44: 55)	43: 55 (37: 62)	49: 49 (40: 59)
Dem: GOP: no-vote (DG)	21.6: 38.9: 38.3 (36.2: 45.3: 17.7)	34.7: 44.4: 19.3 (30.9: 51.8: 16.5)	32.9: 32.9: 32.8 (18.0: 26.5: 55.1)
Dem: GOP: no-vote totals (DG)	9.63m: 17.33m: 17.08m (16.83m: 21.04m: 8.23m)	15.14m: 19.36m: 8.42m (13.37m: 22.40m: 7.14m)	23.71m: 23.71m: 23.62m (12.77m: 18.83m: 39.17m)
Total (w/o 3rd Party)	44.04m (46.10m)	42.92m (42.91m)	71.04m (70.77m)
Dem: GOP: None total Δ (24, 20)	-7.20m: -3.71m: +8.85m	+1.77m: -3.04m: +1.28m	+10.95m: +4.88m: -15.55m
Dem: GOP: None ratio Δ (24, 20)	-14.6: -6.4: +20.6	+3.8: -7.4: +2.8	+14.9: +6.4: -22.3

These results largely verify Finkelstein’s prediction (that the white working class will abandon Harris). They didn’t abandon her to Trump though - they simply didn’t vote.

Black

(Back to Table of Contents)

Black males are 47% of eligible black voters, compared to 53% being black women Pew:

Table IX - 2024 US Presidential Election, black vote breakdown by gender; results reported as 2024-data (2020-data); 2020 turnout: 0.666; 2024 turnout: 0.63; black population: geV: male:female = 47:53

Item	Male	Female
𝑛 2337 (1929)	958 (637)	1379 (1292)
Eligible Population (*)	16.2m (15.1m)	18.3m (17.0m)
Percent Surveyed (PS) (s)	41 (33)	59 (67)
Turnout ({0.495/0.641}*(PS/{47/53}))	43.2 (45.0)	55.1 (81.0)
Dem: GOP of voters (DG)	77: 21 (79: 19)	91: 7 (90: 9)
Dem: GOP: None of all (DGN)	33.3: 9.1: 56.8 (35.6: 8.6: 55.0)	50.1: 3.9: 44.9 (72.9: 7.3: 19.0)
Dem: GOP: None total of all (tDGN)	5.39m: 1.47m: 9.20m (5.37m: 1.29m: 8.31m)	9.18m: 0.71m: 8.22m (12.39m: 1.24m: 3.23m)
Total (w/o 3rd Party)	16.06m (14.97m)	18.11m (16.86m)
Dem: GOP: None total Δ (24, 20)	+0.02m: +0.18m: +0.90m	-3.22m: -0.53m: +4.99m
Dem: GOP: None ratio Δ (24, 20)	-2.3: +0.5: +1.8	-22.8: -3.4: +25.9
2016: Dem: GOP: None total	5.69m: 0.90m: 7.61m	9.13m: 0.39m: 6.92m
Dem: GOP: None ratio Δ (24, 16)	-5.8: +2.9: +4.5	-4.8: +1.6: +3.3

*: Take 0.47/0.53, and multiply it by 34.45m for 2024, and 32.15m for 2020.

This data suggests something interesting: Harris made gains with Black men, albeit marginal (20k votes). Meanwhile Trump made gains (180k votes), still marginal, although many more than Harris. Meanwhile, both Trump and Harris saw declines in the black female population (-3.22m and -0.53m respectively) - the no-vote group actually gained. This is precisely the story Taylor warned about.

Ironically, the very constituency at the heart of the "Harris lost because of racism/misogyny" appears to have actually lost Harris about 3.22m votes (and net over Trump, she "only" lost 2.69m votes), compared to 2020.

Looking at 2016 (Table IV), for black men, Dems lost 0.3m votes, Trump gained 0.57m votes, and no votes gained 1.59m votes. In terms of ratio, Dems have lost 5.8 pp, Trump gained 2.9 pp, and no vote gained 4.5 pp. For black women, Dems gained 0.05m votes, Trump gained 0.32m votes, and no votes gained 1.3m. In terms of ratio, Dems lost 4.8 pp, Trump gained 1.6 pp, and no vote gained 3.3 pp.

Now for income.

Table X - 2024 US Presidential Election, Black vote breakdown by income; results reported as 2024-data (2020-data); 2020 turnout: 0.666; 2024 turnout: 0.63;

Item	< $50k	$50k to $99.999k	$100k+
respondents - 2337 (1929)	1098 (714)	631 (1003)	631 (193)
Percent Surveyed	47% (37%)	27% (52%)	27% (10%)
Percent of Black Population (PBP)	44.4 (46.8)	29.5 (27.8)	26.0 (25.4)
Turnout (PS/PBP)*{0.495/0.641}	52.4 (50.7)	45.3 (119.9)	51.4 (25.2)
Dem: GOP of voters (DG)	91: 6 (86: 13)	69: 30 (95: 4)	83: 15 (NA: NA)
Dem: GOP: no-vote (DG)	47.7: 3.1: 47.6 (43.6: 6.6: 47.6)	31.3: 13.6: 54.7 (113.0: 4.8: -19.0)	42.7: 7.7: 48.6 (NA: NA: NA)
Dem: GOP: no-vote totals (DG)	7.29m: 0.48m: 7.28m (6.56m: 0.99m: 7.16m)	3.18m: 1.38m: 5.56m (10.10m: 0.43m: -1.70m)	3.82m: 0.69m: 4.35m (NA: NA: NA)
Total (w/o 3rd Party)	15.05m (14.71m)	10.12m (8.83m)	8.86m (NA)
Dem: GOP: no-vote totals Δ (24, 20)	+0.73m: -0.51m: +0.12m	-6.93m: +0.96m: +7.26m	NA: NA: NA
Dem: GOP: no-vote ratios Δ (24, 20)	+4.1: -3.5: 0	-81.7: +8.8: +73.7	NA: NA: NA

Looking at the above, a clear anomaly is that the $50k-$99.999k group has a 2020 turnout of 120%. This must be due to some misalignment of the census data and the exit poll data (probably the $50k-$99.999k group is relatively under-reported; not that it is "wrong", but some people may be categorized different by exit pollsters than had been for the Census office). Clearly then, the numbers for 2020 are not exactly real. But they do at least indicate trends.

First, the black working class, while with relatively low turnout, is reliably Democratic. In fact, this is one working class segment in which Harris actually gains voters (500k), in excess of increases in no-vote (120k). Meanwhile, the apparent turnout (whatever the exact value is) of the $50k-$99.999k group collapsed, apparently mostly voters who had voted Democrat in 2020 (No-vote up 7.26m, Harris down 6.93m, 95% of the No-vote, but keep in mind these numbers aren’t quite realistic). We don’t have breakdowns for the $100k+ group for 2020 to judge patterns, but notice their turnout, apparently, doubled (though perhaps some of the "low turnout" in 2020 is an artifact corresponding to the anomalous $50k-$99.999k group).

The issues with the two $50k+ groups makes interpreting them difficult, but certainly, we see consistent black working class turnout (even if not at high rates), which is reliably blue.

Race-Gender, Race-Income Concluding Remarks

(Back to Table of Contents)

Altogether, we can see the results as following:

Figure 1: Top: Changes in total votes for Democrat or GOP or no vote from 2020 to 2024. Bottom: Changes in vote ratios from 2020 to 2024. From Tables VI, VIII, X. See Figure 2 for income bracket white:black:Latino compositions. Notes: For black $50k-$99.999k (see Table X, here with black diagonal lines), the data for 2020 has some issues; probably some mis-alignment between how race+income was recorded by the CNN exit poll, and how the US Census office records it. This may have undercounted black $100k+ voters, for whom there wasn’t enough data for CNN to feel statistically comfortable reporting their vote ratios - hence why no change is reported here for that group (since there is no 2020 reference to compute changes from). Nonetheless, the results are likely indicative of qualitative trends among black $50k-$99.999k voters, and likely actual black $100k+ voter turnout increased. As reported in Table X, black $100k+ voter turnout increased from 25.2% (2020) → 51.4% (2024), consistent with increased turnout among white and Latino $100k+ voters. It seems improbable to me this huge increase is purely an artifact due to issues in recording data about black voters near the $100k threshold between the two black income groups in 2020.

Figure 2: Composition of the income groups for White+Black+Latino

Figure 3: W: White; B: Black; L: Latino; M: Male; F: Female. Top: Change in total votes for Democrat or GOP or no vote from 2020 to 2024. Bottom: Change in vote ratio from 2020 to 2024. From Tables V, VII, IX.

First, from Figure 1, the competitive electorates (those for which no-votes either stagnated, or decreased) were Latinos making $100k+, maybe whites making between $50k-$99.999k, whites making $100k+, and blacks making $100k. The black sub-$50k vote didn’t change much, so it was hardly "competitive", and the black $50k-$99.999k vote likely fell by some large, but uncertain, degree. In other words, this was an election for the rich. Such can be seen in the gender analysis: while the Latino and Black female vote collapsed, the white female vote (presumably disproportionately wealthier) increased. In terms of women, this was an election for white women, leaving POC women - and men of all groups - behind. And overall by race, we saw the 2020→2024 turnout trend as: for whites (66.1% → 68.6%, Δ = 2.5), for blacks (64.1% → 49.5%, Δ = -14.6), and Latinos (63.7% → 51.4%, Δ = 12.3). In other words, this was an election for white people.

Figurex 1

Previous Next

Figure 4: Ratio of votes for {White, Black, Latino} × {Male,Female} voters over 2016, 2020, 2024. From Tables IV, V, VII, IX.

The Youth Vote (18-29) Won’t go Trump, but They’ll sit out the Election

(Back to Table of Contents)

Finkelstein: 18-30 vote won’t go Trump, but will sit out election

This CNN exit poll has 22914 respondants. Among them, 18-29 was 14% (D:R = 54:43); 30-44 was 23% (D:R = 49:48); 45-64 was 35% (D:R = 44:54); and 65+ was 28% (D:R = 49/49). As predicted, 18-29 is the most pro-Harris age group. What’s turnout look like?

In 2024, the US age structure (source; downloaded CSV file here (link will download CSV file) from there on Nov 14 2024) indicates 20.0% of the population is 18-29 This number may not be 100% accurate. The age structure given is in 5 year intervals, so the interval 15-19 overlaps with 18-30. So to compute the total population 18-30, I estimated that 20% of the 15-19 group is 18 and 19; it might be better to linearly fit the changing age structure, and use this to estimate the 18-19 fraction. But 20% is probably close enough , 26.2% is 30-44, 31.0% is 45-64, and 22.8% is 65+. So if the CNN exit poll is representative of national age-based turnout (which will be around 63%), then the turnout for the age groups is 44.1% for 18-29, 55.3% for 30-44, 71.1% for 45-64, and 77.5% for 65+. And thus that for voter groups, the 2024 results broken down by D:R:No Vote are: 18-29 - 23.8:19.0:55.9; 30-44 - 27.1:26.5:44.7; 45-64 - 31.3:38.4:28.9; 65+ - 38.0:38.0:22.5.

Table XI - 2024 US Presidential Election, vote breakdown by age group

Age Group	Dem	GOP	No Vote	Other Vote
18-29	23.8	19.0	55.9	1.3
30-44	27.1	26.5	44.7	1.7
45-64	31.3	38.4	28.9	1.4
65+	38.0	38.0	22.5	1.5

What’s turnout historically? This data from Statista gives the age brackets in 18-24, 25-44, 45-64, and 65+. Here we will look at the turnout per age group for 2008-2020 presidential elections; in parentheses will be adjusted turnouts, as it appears the data under-estimates voter turnout - 2008 (real:data = 61.6/58.2 = 1.06), 2012 (real:data = 58.6/56.5 = 1.04), 2016 (real:data = 60.1/56 = 1.07), and 2020 (real:data = 66.6/61.3 = 1.09). To get 18-29 turnout, I took the weighted average of 18-24 (weight = 7/12) and 25-44 (weight=5/12).

Table XII - turnout by age group, 2008-2024

Year	18-29 (Adjusted)	30-44 (Adjusted)	45-64 (Adjusted)	65+ (Adjusted)
2008	47.5 (50.2)	51.9 (54.9)	65.0 (68.8)	68.1 (72.1)
2012	42.8 (44.4)	49.5 (51.3)	63.4 (65.6)	69.7 (72.3)
2016	43.4 (46.6)	49.0 (52.6)	61.7 (66.2)	68.4 (73.4)
2020	50.9 (55.3)	55.0 (59.8)	65.5 (71.2)	71.9 (78.1)
2024	44.1 (-)	55.3 (-)	71.1 (-)	77.5 (-)

Looking even at the un-adjusted results, 18-29 turnout in 2024 is down from 2020. And the adjusted results show a historic low, and a drop of 11.3 points, or a 20% drop from the 18-29 turnout from 2020.

↑