New England’s demographic makeup, partisan leaning, and historical voting patterns make it an excellent case to analyze to understand the dynamics of the 2024 election.
New England is very liberal. It’s so liberal that the last Republican to carry a state in the region was George W. Bush in 2000, when he beat Al Gore in New Hampshire. While Donald Trump did win Maine’s rural 2nd Congressional District three times, he never won statewide.
New England’s Democratic lean ultimately comes from its demographic composition. New Englanders earn substantially more money than the average American, are much more educated on average, and are much more likely to belong to a union than Americans at large. On top of that, New England is significantly less religious than the nation as a whole. The region has a strong tradition of social liberalism.
However, New England is more liberal than one might expect when considering how white it is. The GOP usually performs better in less racially diverse states. But in the case of New England, the other socio-demographic indicators have made white voters more difficult to reach.
In 2024, Trump’s gains were driven by huge shifts among minority voters, especially Latinos, and strong support from non-college educated white voters. Kamala Harris, on the other hand, fared exceptionally well among whites for a Democratic candidate in a cycle where the nation shifted six points rightward. In fact, Harris performed similarly to Biden with college-educated white voters.
On paper, then, New England should have made it more resistant to conservative inroads last November. Yet the region trended almost seven points rightward from 2020, outpacing the nationwide shift. More interestingly, the biggest shifts came in the bluest strongholds and not in more competitive states.
To unravel the Republican gains, we compared presidential and down-ballot election results across New England from 2024. Our analysis shows that Trump improved significantly with white voters, making the biggest gains with non-college-educated, low-income voters. He also made inroads among the super-rich — those with a median household income exceeding $200K.
Harris, meanwhile, reduced Democratic bleeding with whites. She held on to the most well-educated voters and performed well with the suburban upper-middle class. Down-ballot Democrats running for Senate generally outperformed her with minorities and historically blue constituencies. Harris performed better in ancestrally Republican communities.
Massachusetts
The Bay State, historically the most liberal in the region alongside Vermont, shifted to the right by a staggering 8.3%. Yet Harris ran more than 15 points ahead of Senator Elizabeth Warren, who won her third term in November.
Warren did outperform Harris in urban centers, where Trump made the biggest gains. She also did also quite well in the west of the state, a rural region with lower income. On the other hand, she was annihilated in the Boston suburbs, as Harris’s message resonated more with educated, well-off voters.
As Trump made big inroads across the board, peeling away from Harris also a sizable chunk of rich voters, the correlation between income and 2020->2024 shift was fairly weak, while the relationship between MHI and Harris-Warren difference was remarkable. That is due to Warren performing well in poorer areas of the less-educated southeast, where Harris lost big. Although the former VP did generally better with high-income voters, she struggled with the very rich (>$200,000).
Educational attainment was more polarizing than income. Finstance, Trump was very strong in the southeast, a fairly well-off area with low educational attainment (Plymouth County). In fact, this reflects a broader trend observed in the region and even nationwide; while education usually correlates strongly to income, those voters that didn’t go to college but register high incomes did shift in Trump’s favor.
A U-shape trend line shows how Trump gains were bigger and stronger among both the very rich and low-income voters. It is notable how municipalities’ 2020->2024 rightward shift drops as income increases until around $200,000, before rising again as MHI skyrockets past $250,000.
Vermont
In Vermont, another state where Harris outperformed the senatorial candidate, this time only slightly, the former VP proved her strength with high income voters. She ran either ahead or close to Biden’s numbers from Burlington through Norwich. In comparison, Trump made big gains in rural areas. Again, the correlation between MHI and 2020->2024 shift is not very strong, as educational attainment remains more predictive of electoral movements in presidential races. On the other hand, the correlation between MHI and Harris-Sanders is strong.
This dynamic is epitomized by the area around Manchester, in the south of the state.Hhigh-income with modest educational attainment townships, Trump made gains but Harris still outperformed Sanders. The independent senator ran up big margins in the Trump-dominated, low-income northern Vermont, near the New Hampshire border. He retains significant strength among less engaged voters in more conservative areas that have trended rightward.
A U-shaped trend line shows how Harris struggled with both low- and very-high income voters, while Trump was less effective with upper-middle class, highly-educated whites. As the highest-income townships only approach the $200K threshold, the U is less pronounced.
Connecticut
The same U-shaped trend is visible in Connecticut, where Trump improved significantly in the east of the state and in urban centers home to minority voters. He also made astounding gains in the Gold Coast, a white and very rich region.
A possible explanation as to Trump’s staggering showing in the Gold Coast may include a decisive decrease in turnout among the very rich. Rural municipalities, conversely, registered higher turnout. Trump doing well with the very rich is a pattern common to many regions and cities across the county, from Los Angeles to NYC, results in a very pronounced U-shaped trend line showing the correlation between MHI and 2020->2024 shift.
On the other hand, the correlation between income and Harris-Murphy difference is linear and stronger. Harris was outperformed by the three-term senator in low-income areas in rural eastern Connecticut and in high density areas from New Haven through Hartford. But he underperformed the former Vice President in the Gold Coast, where Trump dramatically improved on his 2020 performance. Murphy retained his appeal with rural voters, but that came to the cost of alienating some well-off, educated voters who backed Harris.
Rhode Island
Rhode Island proves that educational attainment was more descriptive of voting patterns than MHI in New England. The correlation between educational attainment and 2020->2024 was strong (ρ1=-0.82), while considering MHI, the correlation is more moderate (ρ=-0.66).
Interestingly, looking at the difference between Harris and Senator Whitehouse, the correlation increases considering MHI (from ρ=-0.66 to ρ=-0.70) while it remains unchanged taking into account educational attainment, hence reducing the difference between the two indicators. This is a common thread in New England, and underscores how education played a bigger role in the 2020->2024 shift than in the Harris/down-ballot candidate difference.
In this case, the trend line is linear, with richer voters shifting less to the right. This can be attributed to the fact that no township can be classified as “super-rich” – that is to say, with an income of more than $200,000 a year.
Similarly to Connecticut, turnout dropped in the most educated areas. More votes were cast than in 2020 in the northwest. This is a rich but uneducated region that borders some of the Connecticut townships that registered the highest participation.
New Hampshire
In New Hampshire, Donald Trump came within three points of flipping a state Biden carried by seven points in 2020. GOP nominee Kelly Ayotte won the governorship by more than nine points against Democrat Joyce Craig.
Looking at the presidential race, Trump gained significant ground with low-income voters in the southwest and in the rural north. However, he struggled in the richer and more well-educated southeast. Educational attainment was a more predictive indicator than median household income. This was on clear display in the north, where low-income voters with high educational attainment shifted right only marginally.
A similar pattern appears upon comparing Ayotte and Trump. Ayotte outperformed Trump the most in the richest and most well-educated townships in the southeast. Trump did well compared to Ayotte in less well-educated municipalities in the southeast. However, Ayotte outperformed him in low-income, fiscally conservative communities in the north. These areas are more prone to vote conservative in statewide elections.
Educational attainment presents a stronger correlation to the 2020-2024 shift than median household income. This backs up the conclusion that education was the key polarizer in 2024. Relative to 2020, Trump made the biggest gains with less well-educated voters. Harris narrowed the gap relative to Biden in higher-income towns, with the exception of the very rich, with whom Trump did well.
Looking at the Ayotte-Trump gap, the U-shaped line is more pronounced. Ayotte blew the doors off with suburban, educated and rich voters – Haley voters in the primary – only to see her margins over Trump decrease in lower-income towns.
The correlation between the Ayotte-Trump gap and educational attainment is, on the other hand, linear. Ayotte performed better among the most well-educated and struggled more with less well-educated voters. Notably, Ayotte outperformed the most where Trump gained the least, while the townships that registered the biggest rightward shifts at the top of the ticket saw a smaller Ayotte-Trump gap.
Western municipalities that overwhelmingly supported Bernie Sanders in the 2016 Democratic primary have registered the largest 2020->2024 rightward shifts. At the same time, Harris had the slimmest outperformance over Craig in those townships. This trend shows Trump’s strengths among former democratic constituents, who may be less prone to support more traditional, down-ballot republicans.
Maine
The same patterns hold in Maine. Trump did better with rural and less rich voters, Harris in the educated suburbs. Educational attainment levels better describe the electorate when it comes to 2020->2024 shift. However, the gap shrinks looking at the difference between Harris and Senator Angus King, who coasted to reelection running as an independent who caucuses with democrats.
Interestingly, Trump made bigger-than-expected gains in the Portland regions, home to a significant share of Latino population. Harris performed quite well along the coast, up to Bar Harbor. That is why the liberal 1st Congressional District shifted almost as much as the more rural 2nd Congressional District (1.5% and 1.8%); Harris was able to limit her losses among educated whites in Jared Goldan’s district. The more rural second district registered around 20,000 more votes compared to 2020, while the liberal first district saw a 5,000-vote drop, confirming how low enthusiasm was among more educated, well-off voters in New England.
Given that the state has a lower income than its New England neighbors, there’s no U-shape when it comes to correlation trend lines.
Unlike other states in New England, the correlation between Trump gains from 2024 and Median Household Income is perfectly linear. This means Harris did better even with the very rich, a group that trended rightward in states like Connecticut and Massachusetts. This could be ascribed to the fact that Maine whites registered the smallest shift in Trump’s favor.
It’s also worth looking at the performance of Senator Susan Collins, who won re-election in 2020. She outperformed Trump by 18 points. Next year, she will be defending a seat that Democrats consider key to winning back control of the chamber.
Unsurprisingly, the towns where Collins outperformed Trump the most in 2020 are in high-income, educated areas that tend to vote Democratic in presidential elections. Importantly, these areas have proven to turn out at higher rates in low-turnout elections. The charts show how Trump had to bank on low-propensity voters in 2024. On the other hand, Collins was able to gather support even from more engaged voters. These voters make up a larger portion of the electorate, hence positioning her in a sustainable position ahead of 2026.
1ρ denotes Spearman’s rank correlation coefficient. When ρ is closer to 1, the correlation is stronger.
