Data Analysis

Trump:
The Machine Politician

Whether you see it as delivering for his voters or telling everyone what they wanted to hear, Trump ran effectively unopposed in his outreach to communities that vote in blocs. Biden was near death's door. Harris never showed up.

The single strongest predictor of how a county swung in 2024 wasn't education. Wasn't race. Wasn't rural vs urban.

It was language.

Non-English speaking communities of all stripes—Hispanic, Arab, Asian, Eastern European—swung massively toward Trump. The correlation is striking: r = -0.77, weighted by population. Look at those maps. The overlap is uncanny.

High Non-English CVAP Counties

Counties with the highest non-English speaking citizen voting age populations — showing the language-swing correlation

CountyNon-Eng%2020202420→24Community
Webb County (Laredo)(TX)89.0%D+23.3R+2.2-25.4Hispanic border
Hidalgo County(TX)80.7%D+17.1R+2.9-19.9Hispanic border
Imperial County(CA)72.1%D+24.4R+0.9-25.2Hispanic agricultural
Miami-Dade County(FL)70.9%D+7.3R+11.4-18.8Cuban/Latino
Cameron County(TX)70.2%D+13.2R+5.8-18.9Hispanic border
El Paso County(TX)68.1%D+35.1D+15.1-20.0Hispanic border
Bronx County(NY)53.1%D+67.5D+44.9-22.6Hispanic/Black working class
Osceola County(FL)51.1%D+13.8R+1.4-15.3Puerto Rican
Hudson County(NJ)48.8%D+46.6D+27.9-18.7Hispanic + diverse
Los Angeles County(CA)48.1%D+44.0D+32.9-11.1Diverse metro
Queens County(NY)47.9%D+45.2D+24.1-21.1Diverse immigrants
Yuma County(AZ)47.5%R+6.2R+20.4-14.2Hispanic border
Passaic County (Paterson)(NJ)41.7%D+16.7R+2.9-19.5Arab + diverse
Brooklyn (Kings)(NY)38.3%D+54.8D+43.0-11.8Diverse urban
San Bernardino County(CA)37.6%D+10.7R+2.1-12.8Diverse working class
Rockland County(NY)35.0%D+1.7R+11.7-13.4Orthodox Jewish

Every county above 35% non-English CVAP swung toward Trump. The higher the non-English share, the larger the swing. This is the core finding: language predicts swing better than any other single variable.

CVAP Non-English: Share of Citizen Voting Age Population speaking a language other than English at home (Census ACS 2024 5-year estimates). All margins and swings from Akashic Edge election database.

The Full Model: 14 Variables

Beyond CVAP non-English %, our model captures 13 other demographic and political factors. Each tells its own story about how Trump assembled his coalition. Click any card to see detailed statistics and interpretation.

Model Summary

Weighted Least Squares • CVAP weights • 2024 ACS 5-Year

3,143
Counties
77.1%
R² (Variance Explained)
751.8
F-statistic
14
Predictors
Adj. R²
0.7699
P(F-stat)
<0.001
AIC
17666
Durbin-Watson
1.92
#1
🗣️

CVAP Non-English %

R-shiftβ = -0.65R² = 61% alone
← R-shiftStandardized Effect (β)D-shift →
Coef
-0.2084
Std Err
0.0042
t-stat
-49.7
P-value
<0.001
95% Confidence Interval[-0.217, -0.200]
R-shift
D-shift

The dominant predictor—and it's not close. Citizens who speak a non-English language at home swung dramatically toward Trump across every community: Hispanic, Asian, Arab, and other immigrant-origin groups. This single variable explains 61% of all county-level swing variance.

Interpretation

For every 1 percentage point higher non-English CVAP, counties swung 0.21 points more Republican. The 95% confidence interval is extremely tight [−0.217, −0.200], reflecting exceptional statistical certainty.

Exemplar Counties
D+5.0R+16.0
R+21.0
R+60.7R+65.4
R+4.7
Source: ACS 2020-2024 5-Year (CVAP Special Tabulation)
#2
💻

Work From Home %

D-shiftβ = 0.39
← R-shiftStandardized Effect (β)D-shift →
Coef
+0.2552
Std Err
0.0087
t-stat
29.4
P-value
<0.001
95% Confidence Interval[0.238, 0.272]
R-shift
D-shift

The 'laptop class' effect. Remote workers in the knowledge economy—concentrated in affluent metros and tech hubs—swung toward Democrats, partially offsetting working-class losses elsewhere.

Interpretation

For every 10 percentage points higher work-from-home rate, counties swung 2.6 points more Democratic. The professional class diverged sharply from working-class voters.

Exemplar Counties
D+63.5D+58.1
R+5.5
R+67.9R+71.7
R+3.8
Source: ACS 2020-2024 5-Year (Commuting Characteristics)
#3

Evangelical %

D-shiftβ = 0.27
← R-shiftStandardized Effect (β)D-shift →
Coef
+0.0497
Std Err
0.0050
t-stat
9.9
P-value
<0.001
95% Confidence Interval[0.040, 0.060]
R-shift
D-shift

High-evangelical counties are dry, conservative America—places that were already voting Republican at 60-70% margins. Their swings were modest because there simply weren't many voters left to convert. Low-evangelical areas (like Mormon Utah or secular New England) had more room to move in either direction.

#4
🏛️

LDS (Mormon) %

D-shiftβ = 0.15
← R-shiftStandardized Effect (β)D-shift →
Coef
+0.0642
Std Err
0.0065
t-stat
9.9
P-value
<0.001
95% Confidence Interval[0.051, 0.077]
R-shift
D-shift

Mormon country resisted Trump. High-LDS counties swung less Republican than demographically similar areas—lingering institutional distaste from 2016 and Mitt Romney's continued influence on Utah politics.

#5
🍺

German Ancestry %

D-shiftβ = 0.15
← R-shiftStandardized Effect (β)D-shift →
Coef
+0.0422
Std Err
0.0066
t-stat
6.4
P-value
<0.001
95% Confidence Interval[0.029, 0.055]
R-shift
D-shift

German-American counties—concentrated in Wisconsin, Minnesota, and the northern Great Plains—resisted the national swing toward Trump. The upper Midwest's progressive tradition held firm.

#6
✝️

Black Protestant %

R-shiftβ = -0.13
← R-shiftStandardized Effect (β)D-shift →
Coef
-0.1195
Std Err
0.0146
t-stat
-8.2
P-value
<0.001
95% Confidence Interval[-0.148, -0.091]
R-shift
D-shift

Trump made inroads in Black church communities. Counties with higher Black Protestant adherent rates saw larger Republican swings—reflecting both persuasion among Black men and differential turnout patterns.

#7
🎯

Swing State

D-shiftβ = 0.13
← R-shiftStandardized Effect (β)D-shift →
Coef
+1.1458
Std Err
0.1024
t-stat
11.2
P-value
<0.001
95% Confidence Interval[0.945, 1.347]
R-shift
D-shift

Campaign investment matters. Counties in the seven swing states (AZ, GA, MI, NC, NV, PA, WI) performed 1.1 points more Democratic than equivalent counties elsewhere. Media saturation and field operations moved votes.

#8
🍝

Italian Ancestry %

R-shiftβ = -0.11
← R-shiftStandardized Effect (β)D-shift →
Coef
-0.1257
Std Err
0.0128
t-stat
-9.8
P-value
<0.001
95% Confidence Interval[-0.151, -0.101]
R-shift
D-shift

Italian-Americans continued their multi-decade drift from the Democratic Party. Staten Island—22.9% Italian—delivered Trump's largest swing in any NYC borough. Working-class Catholic ethnics have realigned.

#9
🌻

Ukrainian Ancestry %

D-shiftβ = 0.10
← R-shiftStandardized Effect (β)D-shift →
Coef
+0.9500
Std Err
0.1530
t-stat
6.2
P-value
<0.001
95% Confidence Interval[0.650, 1.250]
R-shift
D-shift

Ukrainian-Americans rewarded Biden's support for Ukraine against Russian aggression. The strongest pro-Democratic ancestry effect in the model—each percentage point of Ukrainian ancestry correlated with nearly a full point more Democratic swing.

#10
🚚

Migration Effect

Variesβ = 0.08
← R-shiftStandardized Effect (β)D-shift →
Coef
+0.0125
Std Err
0.0011
t-stat
11.4
P-value
<0.001
95% Confidence Interval[0.010, 0.015]
R-shift
D-shift

Migration reshapes electorates. Based on NYT analysis of interstate movers' partisan lean. Counties receiving Democratic-leaning movers (Atlanta suburbs) held up; counties receiving Republican-leaning movers (Idaho exurbs) swung harder right.

#11
🕌

Arab Ancestry %

R-shiftβ = -0.08
← R-shiftStandardized Effect (β)D-shift →
Coef
-0.7197
Std Err
0.0660
t-stat
-10.9
P-value
<0.001
95% Confidence Interval[-0.849, -0.590]
R-shift
D-shift

Gaza broke the Arab-American vote. Biden's handling of the Israel-Gaza conflict drove a historic rupture. Dearborn and Paterson—centers of Arab-American life—saw some of the largest swings in the country.

#12
✡️

Jewish Population %

R-shiftβ = -0.08
← R-shiftStandardized Effect (β)D-shift →
Coef
-0.2435
Std Err
0.0167
t-stat
-14.5
P-value
<0.001
95% Confidence Interval[-0.276, -0.211]
R-shift
D-shift

Jewish communities swung toward Trump across the board. October 7th and rising antisemitism fears drove a historic shift. Brooklyn (19% Jewish) and Rockland County (18.9% Jewish) both saw double-digit Republican swings.

#13
🪷

Indian Ancestry %

R-shiftβ = -0.05
← R-shiftStandardized Effect (β)D-shift →
Coef
-0.1813
Std Err
0.0259
t-stat
-7.0
P-value
<0.001
95% Confidence Interval[-0.232, -0.130]
R-shift
D-shift

South Asian Americans shifted right. Counties with higher Indian-ancestry populations swung toward Trump, part of a broader pattern of immigrant-origin communities moving Republican as they assimilate economically.

#14

Portuguese Ancestry %

R-shiftβ = -0.03
← R-shiftStandardized Effect (β)D-shift →
Coef
-0.1537
Std Err
0.0356
t-stat
-4.3
P-value
<0.001
95% Confidence Interval[-0.224, -0.084]
R-shift
D-shift

Portuguese-American communities in southeastern Massachusetts and California's Central Valley shifted right. Working-class Catholic voters following the broader ethnic Catholic realignment.

All coefficients significant at p < 0.001. Model weighted by CVAP (Citizen Voting Age Population).

Outcome: Swing = margin_2024 − margin_2020 (positive = D-shift, negative = R-shift)

Case Study

Dearborn, Michigan

Dearborn, Michigan (Arab American)

Largest Arab American concentration in the US — Gaza backlash in 2024

2012 → 2024 Swing
Loading map...
YearDem %Rep %3rd PartyMarginSwing
201266.6%32.3%1.1%D+34.3
201663%30.7%6.3%D+32.3-2.0
202068.8%29.7%1.5%D+39.1+6.8
202436.3%42.5%21.3%R+6.2-45.3
Dearborn City
41,767 votes (2024)
Stein: 18.4% (7,702 votes)D+39 → R+6-45.3

The single biggest city-level swing in America. Dearborn went from D+39 in 2020 to R+6 in 2024 — a 45-point collapse driven by Gaza. Jill Stein captured 18.4% of the vote (7,702 ballots), nearly all from would-be Harris voters. The Arab American community sent the clearest possible message: foreign policy matters to diaspora communities.

Trump's play: promise everything to everyone. Pro-Israel donors got their messaging. Arab voters got theirs. The bet was that these audiences wouldn't see each other's ads.

It worked.

The Michigan math:

Trump won Michigan by 80,106 votes. In 2020, Biden carried Dearborn by 17,480 net votes. In 2024, Trump won it by 2,594.

That's a 20,074 vote swing from a single city. One-quarter of Michigan's entire margin.

The Harris campaign's response to Gaza? Essentially nothing. No counteroffer. No outreach. Just silence while votes walked out the door—and Michigan with it.

Case Study

Laredo, Texas

In 2012, the Laredo metro was Obama's second-best in the entire country—trailing only Santa Cruz, California. In 2024, it voted for Trump.

Webb County (Laredo), TX

95% Hispanic border county — Obama's #2 metro in 2012

YearDem %Rep %Margin
201276.4%22.5%D+53.9
202448.5%50.6%R+2.1

R+56 swing in 12 years. From Obama's second-best metro to voting Republican.

Webb County is 95% Hispanic. This isn't a story about white voters. It's not about rural resentment. It's about a community that gave Obama his second-best metro performance in America—and flipped to Trump twelve years later.

The border wasn't an abstraction here. It was daily life. And somewhere along the way, the Democratic coalition lost the plot.

From Obama's second-best metro to voting Republican. A 56-point swing in twelve years.

Case Study

Lake & Peninsula Borough, Alaska

Lake and Peninsula Borough, Alaska

Home to Yup'ik Native Alaskans with Russian Orthodox heritage

2012 → 2024 Swing
Loading map...
CountyNon-Eng%201220162020202412→1616→2020→24
Lake and Peninsula Borough(AK)19.9%D+17.6D+5.3D+12.7R+10.4-12.3+7.4-23.1

From D+13 in 2020 to R+10 in 2024 — a 23-point swing. This tiny borough (pop. 1,476) is home to Yup'ik Native Alaskans whose ancestors intermarried with Russian Orthodox missionaries in the 1800s. Shows how micro-coalition targeting can flip even the most obscure communities.

Mostly Yup'ik Native Alaskans whose ancestors intermarried with Russian Orthodox missionaries in the 1800s. They've been Orthodox Christians for generations.

On October 29, 2024, Trump did something no major presidential candidate had done before: he addressed Coptic Christians by name. One tweet. That's all it took.

The Orthodox world noticed. The Coptic bishops issued a statement congratulating him after the election. And in remote Alaska, Yup'ik communities who share the Orthodox faith swung nearly 30 points.

A 30-point swing from a single tweet. That's the return on investment of targeted micro-coalition outreach.

Small community. One tweet. R+30 swing. That's machine politics.

Case Study

The Lumbee Tribe

Robeson County, North Carolina. Home of the Lumbee Tribe—55,000 members, the largest tribe east of the Mississippi, fighting for full federal recognition since 1888.

Robeson County, NC (Lumbee Tribe)

Home of the Lumbee — largest Native American tribe east of the Mississippi

2012 → 2024 Swing
Loading map...
CountyNon-Eng%201220162020202412→1616→2020→24
Robeson County(NC)8.2%D+17.4R+4.3R+18.6R+27.6-21.7-14.3-9.0

D+17 in 2012 → R+28 in 2024 — a 45-point swing. Trump signed the Lumbee Recognition Act in December 2024, granting full federal tribal status after 100+ years of denial. Like Lake and Peninsula (Greek Orthodox), targeted policy delivery converted a Democratic-leaning community into a Republican one.

Trump promised federal recognition. On January 28, 2025, he signed a presidential memo directing Interior to advance recognition. On December 18, 2025, he signed the Lumbee Fairness Act into law. The Lumbee became the 575th federally recognized tribe.

137 years of waiting. Democrats had decades to deliver. They didn't.

The Clinton Echo

History rhymes—and these communities remember

Population-Weighted Correlation
r = -0.51
Clinton's 1992→1996 swing vs Trump's 2020→2024 swing
n = 3,133 counties
📈

A strong inverse relationship. Counties where Bill Clinton improved most in 1996 are the same counties that swung hardest to Trump in 2024. What does that mean?

Three Eras, Four Counties

Clinton's 1996 gains vs Trump's 2024 gains — the same communities, mirror-image swings

County19921996Δ20202024Δ
Webb County (Laredo), TXD+27D+58+31D+23R+2-26
Queens, NYD+35D+52+17D+45D+24-21
Miami-Dade, FLD+4D+20+16D+7R+11-19
Bethel Census Area, AKD+0D+27+27D+28D+12-16
NationalD+5.6D+8.5+3D+4.5R+1.5-6

The same communities that responded to Clinton's 1996 message responded to Trump's version of it in 2024. The realignment Clinton started is completing — just not in the direction Democrats hoped.

🎯

Clinton's 1996 Formula

  • Signed the Illegal Immigration Reform Act
  • Passed welfare reform ("end welfare as we know it")
  • "The era of big government is over"
  • Ran to center on cultural issues
Result
Consolidated immigrant working class
🔄

Trump's 2024 Reversal

  • Border security as central message
  • Direct outreach to Hispanic working class
  • Economic populism, anti-inflation focus
  • Targeted ethnic/religious bloc appeals
Result
Captured immigrant working class

The same communities that responded to Clinton's 1996 message responded to Trump's version of it in 2024. Twenty-eight years later, the realignment Clinton started is completing—just not in the direction Democrats hoped.

Why Trump Ran Unopposed

The real story isn't inflation—it's absence

The standard explanation for 2024 is "inflation" or "vibes" or "incumbency drag." Those matter. But they don't explain why the swing was so concentrated in non-English-speaking communities and ethnic/religious blocs.

The real story: Trump practiced politics, and Democrats didn't.

Since 1972, when Democrats reformed their primary process to eliminate party bosses, they became allergic to transactional coalition-building—the kind of politics where you make a specific promise to a specific group and deliver. The kind of politics that built the New Deal coalition.

Modern Democrats think that's corruption. They think "good policy" should speak for itself. They staff campaigns with credentialed millennials who say "Latinx" to Hispanic voters who've never heard the word. They lecture instead of listen.

Biden's 2024 collapse wasn't just about age. It was a White House full of yes-men who couldn't tell him the truth, surrounded by staffers who thought posting on social media was the same as organizing.

Harris inherited this mess and ran a bizarrely low-energy campaign. No outreach to Arab voters on Gaza. No counter-programming to Trump's ethnic coalition plays. Just... nothing.

Trump filled the vacuum. He showed up. He made promises. He treated these communities like their votes were worth fighting for.

That's not genius. That's basic politics. But when your opponent isn't playing, basic is enough.

The Ephemeral Coalition?

2025 suggests the swing might not stick

One caveat: this swing might not stick.

Non-English-speaking immigrants are disproportionately price-sensitive. They don't have deep ideological commitments to either party. They voted their wallets in 2024, and their wallets were hurting.

We're already seeing the snapback.

In the November 2025 governor elections, Democrats recovered sharply in exactly the communities that swung hardest to Trump:

The 2025 Recovery

November 2025 governor results in immigrant-heavy counties — snap back from Trump surge

CountyState202020242025Recovery
Hudson CountyNJD+46.6D+27.9D+51.2
+23.3+4.6 vs '20
Manassas Park cityVAD+33.1D+19.9D+42.5
+22.6+9.4 vs '20
Passaic CountyNJD+16.7R+2.9D+16.3
+19.298% back
Prince William CountyVAD+27.0D+17.9D+34.4
+16.5+7.4 vs '20
Fairfax CountyVAD+41.9D+34.7D+48.0
+13.3+6.1 vs '20

↑ = exceeded 2020 levels. Passaic County went from D+16.7 (2020) to R+2.9 (2024) to D+16.3 (2025) — a complete reversal in twelve months.

Passaic County—home to Paterson's Arab community—went from D+16.7 (2020) to R+2.9 (2024) to D+16.3 (2025). Nearly the exact same margin as 2020. A complete reversal in twelve months.

"Hispanics married President Trump, but they're only dating the GOP."

— Rep. Maria Elvira Salazar (R-FL), November 2025

The 2024 coalition was a rental, not a realignment. Trump delivered a masterclass in transactional politics. But transactional coalitions require continuous delivery. The question is whether Democrats can remember how the game works—or whether they'll keep losing to an opponent who actually shows up.

If you found this analysis useful, share it:

Join the Discussion

Have thoughts on this analysis? Questions about the data? Join our Discord community to discuss, debate, and dive deeper.

Join Discussion on Discord

Free to join • Real-time discussion • No account required (but recommended)

Data: Akashic Edge 14-variable WLS regression (n=3,143 counties, R²=0.77, weight=CVAP). Full model output in backend/scripts/regression-2024.py. Dearborn city results from Wayne County election records. Alaska 2020 borough estimates from RRH Elections; Alaska 2024 borough estimates via proportional allocation of non-ED votes. 2025 governor results from Virginia and New Jersey Secretaries of State.

Introduction
1 / 9

Loading National Swing map...