There is a durable genre of political feature writing built on geography: the Democrat holding out in coal country, the Republican adrift in Brooklyn, the idea that where Americans live quietly remakes what they believe. The Cooperative Election Study — with 3,436,567 policy-item responses between 2006 and 2021, each respondent matched to their county’s presidential result — lets you test that idea at scale. It mostly fails.
Take twelve policy questions the CES asked repeatedly — guns, immigration, abortion, health care, climate, gay marriage — and compare people across the starkest geographic divide available: counties a major party carried by at least 20 points. Within the same kind of deep-red or deep-blue county, Republicans and Democrats sit an average of 41.9 percentage points apart. Move a partisan from one deep county to its opposite and their party’s average position shifts by just 5.6 points. Across the twelve issues, the median ratio of the party gap to the place gap is 7.8 to one.
5.6 points — average deep-blue vs. deep-red county gap inside the same party.
The ledger, issue by issue
The chart below is the whole argument in one frame. Each row is an issue; each pair of connected dots is one party, spanning its supporters in Trump+20 counties (hollow) and Biden+20 counties (solid). The short colored connectors are everything that “place” buys you. The white space between the red and blue pairs is party.
Party is the chasm; place is a rounding error
WEIGHTED % SUPPORTING EACH POLICY, BY PARTY × DEEP COUNTY TYPE — CES POOLED 2013–2021. HOVER ANY DOT FOR VALUES AND SAMPLE SIZES.
Hollow = residents of Trump+20 counties; solid = Biden+20 counties. Right gutter: average party gap ÷ average place gap. Party ID is 3-point; leaners excluded. Every cell has at least 6,824 respondents.
The extremes deserve to be read slowly. On repealing the Affordable Care Act, Democrats in deep-red counties (29.9 percent support) and deep-blue counties (29.3 percent) are 0.6 points apart — statistically, the same people. Republicans living in those same deep-red counties, often on the same streets, support repeal at 75.5 percent: a 45.5-point party gap, and a symmetric party-to-place ratio of 18. Computed against the Democratic place gap alone, the ratio is 73.
The border wall is nearly as lopsided. In deep-red counties, 13.9 percent of Democrats and 80.4 percent of Republicans backed wall spending — a 66.5-point gulf between neighbors — while Democrats across the full red-to-blue county spectrum differ from each other by 2.0 points. Symmetric ratio: 15; against the Democratic place gap alone, 34. Whatever the wall debate was, it was not a conversation between regions. It was a conversation between parties, conducted identically everywhere.
An obvious objection: maybe red-county Democrats just look like blue-county Democrats on average because of who they are — whiter, older, less college-educated — and adjusting for composition would reveal a hidden local imprint. It does not. Reweighting via a demographically adjusted linear probability model (education, race, age, gender), the Democratic place gap on an assault-weapons ban goes from 5.0 points raw to 6.0 points adjusted (standard error 0.6). There is no suppressed geography effect waiting underneath the demographics. There is simply very little within-party variation for geography to explain.
The vanishing local Democrat
It was not always quite this flat. Track the same items across presidential cycles and you can watch nationalization happen in real time: the line between the parties climbs while the line between a party’s red-county and blue-county wings sinks toward zero.
Party gaps rose; place gaps fell toward zero
ABSOLUTE GAPS IN WEIGHTED % SUPPORT, BY PRESIDENTIAL CYCLE (PRES_YEAR_USED). RED/BLUE COUNTIES = TWO-PARTY MARGIN ±5PP. HOVER POINTS FOR VALUES.
2004-cycle estimates come from the 2007 CES wave only and carry real sampling noise (minimum cell N 734). Background checks, lower right, is the deliberate outlier — see section 03.
The cleanest case is a path to legal status for undocumented immigrants. In the 2004 cycle, blue-county Democrats were 10.1 points more supportive than red-county Democrats — a real, visible local difference. That place gap then fell monotonically: 10.1 to 6.5 to 5.3 to 1.0 to 1.3 by the 2020 cycle. Over the same span the gap between the parties grew from 27.0 to 44.5 points, with Democrats moving from 47.8 to 87.9 percent support and Republicans from 20.7 to 43.4 percent. Both parties moved left on the question; Democrats moved much further, and their geographic wings fused along the way.
The partisan gaps ballooned almost everywhere. ACA repeal: from 20.8 points in the 2012 cycle to 56.9 in 2020, a rise of 36.1 points — partly because some early-2010s Democrats opposed the law from the left, before repeal politics sharpened its meaning. Border security: 27.6 to 48.1, up 20.5. Always-allow abortion: 43.7 to 54.8, up 11.2. An assault-weapons ban: 41.1 to 52.5, up 11.4. The local Democrat did not move to a swing state. The category dissolved.
The exception that proves the rule
One issue refuses the script: universal background checks. Its partisan gap actually shrank, from 15.3 points in the 2012 cycle to 13.2 in 2020, and Republican support never drops below 80.5 percent in any county bin from Trump+20 to Biden+20. It is the one item in the set on which America is sorted by neither party nor place — a reminder of what consensus looks like in this data, and how rare it is.
Background checks: the unsorted issue
WEIGHTED % SUPPORTING UNIVERSAL BACKGROUND CHECKS BY PARTY AND COUNTY ENVIRONMENT, CES POOLED 2013–2021. HOVER POINTS FOR VALUES.
Pure Independents (leaners excluded) are the only group where county environment registers: a 7.4-point place gap, 80.8% in deep-red counties vs 88.3% in deep-blue.
The fine print is telling. Among partisans, place does nothing here. Among pure Independents — the people with no party anchor — support runs from 80.8 percent in deep-red counties to 88.3 percent in deep-blue ones, a 7.4-point gradient. Where party identity is absent, local environment gets a little purchase. Where it is present, it doesn’t.
Where place still works: blue-county Republicans
The 5.6-point average hides one systematic pattern, and it inverts the usual human-interest framing. The partisans whose views still bend with their surroundings are not red-county Democrats. They are blue-county Republicans, and the bending is specifically cultural.
The local imprint is Republican — and cultural
DEEP-BLUE MINUS DEEP-RED COUNTY GAP IN % SUPPORT, WITHIN PARTY, AFTER DEMOGRAPHICS + STATE FIXED EFFECTS. WHISKERS = ±2 STATE-CLUSTERED SE. HOVER BARS FOR DETAIL.
Weighted LPM among deep-county partisans; controls: education, race, age, gender; state fixed effects; SEs clustered by state. CES pooled 2013–2021.
Raw, the Republican gaps are roughly twice the Democratic ones. On an assault-weapons ban, Republicans in Biden+20 counties support the ban at 47.9 percent versus 37.5 percent in Trump+20 counties — a 10.4-point spread, against 5.0 for Democrats. On always allowing abortion: 38.1 versus 23.6 percent, a 14.5-point Republican spread, against 9.7 for Democrats. On legalizing gay marriage: 40.2 versus 27.3, a 12.9-point spread, against 8.0.
And it is not composition, and not just blue states. With demographic controls plus state fixed effects — comparing deep-blue-county and deep-red-county Republicans within the same state — the Republican gaps remain 8.9 points on the weapons ban (SE 1.0), 8.2 on abortion (SE 1.1), and 8.9 on gay marriage (SE 2.1). The matching Democratic figures are roughly half: 4.7 (SE 1.2), 6.5 (SE 1.6), and 6.4 (SE 2.3). On economic items the imprint vanishes for everyone once adjusted — Medicare for All shrinks to 2.6 points for Republicans and 1.6 for Democrats; the border wall flips slightly negative, at 2.2 and 0.8 points below zero. The Republican gun gap is also no Trump-era artifact: cycle by cycle from 2012 to 2020 it holds at 9.6–11.5 points under the deep-county definition and 6.8–8.0 under the looser ±5-point one.
Part of the asymmetry is mechanical: Democrats sit near ceiling-level consensus on several culture items, leaving little room to vary. But abortion — where Democrats are well short of unanimity — shows the same two-to-one pattern, so the ceiling cannot be the whole story. If you want to find a partisan whose neighbors still seem to matter, look for a Republican in a deep-blue metro. The electoral footnote writes itself: the geographically persuadable Republicans live in counties Democrats already win.
What this does and doesn’t show
None of this is causal. County color is descriptive context: people choose where to live, and the residual 5.6-point place gap could be weak contextual influence, self-selection of unusual partisans into mismatched places, or both. The state-fixed-effects models rule out composition on measured demographics; they cannot rule out sorting on unmeasured taste. What the data do establish is an upper bound: even reading every place difference as pure local influence, geography moves a partisan’s issue positions by single digits while party membership predicts gaps averaging 41.9 points — in continuous terms, within-party support shifts only 0.3 to 1.9 points per 10-point change in county presidential margin. Place may shape turnout, candidates, and the texture of political life. On what partisans actually believe about policy, the county line has almost nothing left to say.