Behind enemy lines, county by county

Weighted share of each party's local partisans living in a county their party lost · pooled 2020+2024 CES cycles, each respondent scored against their own cycle's result · hover a county for detail

Stranded Democrats

Share of local Democrats in a county the party lost

Stranded Republicans

Share of local Republicans in a county the party lost

579 counties with N≥50 pooled partisan respondents (2020+2024 CES cycles, weight_cumulative). Beige counties have too few respondents and are suppressed; Alaska is excluded. Mappable counties skew urban: they cover 85.7% of weighted Democrats but only 71.4% of weighted Republicans, so deep-red rural America is under-shown here — the state and region figures below fill that gap.

Where each party's stranded actually live

Top 10 states by share of each party's stranded population, pooled 2020+2024 cycles · the notch marks that state's share of all of the party's partisans · hover for detail

Share of all stranded Democrats

Share of all stranded Republicans

Bars are weighted shares of each party's stranded population, not shares of state populations. A bar that extends past its notch means the state holds more of the party's stranded members than of its members overall — Missouri and Indiana over-index for Democrats, California for Republicans.

The stranded gap, 2004–2024 cycles

Weighted share of each party living in a county their party lost, by presidential cycle · hover for values

CES cumulative file, weight_cumulative. Each cycle pools the survey waves matched to that presidential result (2020 cycle = 2020–2023 waves); the 2024 cycle is a single survey wave. Solid series uses pid3 partisans only; the toggle assigns leaners to their party.

Landslide counties vs landslide voters, 1972–2024

Counties with a 20+ two-party-point margin: share of all counties vs share of all two-party votes cast in them · hover for values

County presidential returns, 1972–2024, Alaska excluded; vote shares weighted by two-party votes. 1972 is a Nixon-blowout outlier (58.2% of votes in landslide counties), so the trend reading anchors on 1976. The widening gap between the lines is the story: landslide counties multiplied much faster than landslide voters.