{
  "alt": "Line chart of the correlation of five county characteristics with the county two-party margin in each presidential election from 2000 to 2024, in five views. Vote-weighted, the white share is the most negative through 2016, college share rises from {fact:r_edu_2000_w} to {fact:r_edu_2024_w}, income rises from {fact:r_inc_2000_w} to {fact:r_inc_2024_w}, and median age moves from {fact:r_age_2000_w} to {fact:r_age_2024_w}. In 2024 college share ({fact:r_edu_2024_w}), rural share ({fact:r_rural_2024_w}) and the white share ({fact:r_race_2024_w}) are the three largest in size. Two further views plot the share of resamples of states in which each axis has the largest absolute correlation.",
  "caption": "The white share has the largest absolute correlation in the point estimates through 2016, and in 2020 and 2024 the leading axis is not separated from the next: in 2024 college share leads rural share by {fact:gap_2024_w}, an interval that includes zero. Each line is the correlation across counties between one characteristic and the county two-party margin, positive when counties with more of the characteristic voted more Democratic. Bands are 95% intervals with errors clustered by state, drawn for education, race and ethnicity and rurality only.",
  "data": [
    {
      "axis": "Education",
      "band_high": 0.381,
      "band_low": 0.182,
      "conf_high": 0.381,
      "conf_low": 0.182,
      "estimate": 0.282,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "College share",
      "n_units": 3113,
      "weighting": "Vote-weighted",
      "year": 2000
    },
    {
      "axis": "Education",
      "band_high": 0.472,
      "band_low": 0.284,
      "conf_high": 0.472,
      "conf_low": 0.284,
      "estimate": 0.378,
      "first_high": null,
      "first_low": null,
      "first_share": 1.5,
      "measure": "College share",
      "n_units": 3113,
      "weighting": "Vote-weighted",
      "year": 2004
    },
    {
      "axis": "Education",
      "band_high": 0.511,
      "band_low": 0.343,
      "conf_high": 0.511,
      "conf_low": 0.343,
      "estimate": 0.427,
      "first_high": null,
      "first_low": null,
      "first_share": 0.5,
      "measure": "College share",
      "n_units": 3113,
      "weighting": "Vote-weighted",
      "year": 2008
    },
    {
      "axis": "Education",
      "band_high": 0.488,
      "band_low": 0.342,
      "conf_high": 0.488,
      "conf_low": 0.342,
      "estimate": 0.415,
      "first_high": null,
      "first_low": null,
      "first_share": 0.1,
      "measure": "College share",
      "n_units": 3112,
      "weighting": "Vote-weighted",
      "year": 2012
    },
    {
      "axis": "Education",
      "band_high": 0.64,
      "band_low": 0.514,
      "conf_high": 0.64,
      "conf_low": 0.514,
      "estimate": 0.577,
      "first_high": null,
      "first_low": null,
      "first_share": 0.9,
      "measure": "College share",
      "n_units": 3111,
      "weighting": "Vote-weighted",
      "year": 2016
    },
    {
      "axis": "Education",
      "band_high": 0.696,
      "band_low": 0.579,
      "conf_high": 0.696,
      "conf_low": 0.579,
      "estimate": 0.637,
      "first_high": null,
      "first_low": null,
      "first_share": 17.3,
      "measure": "College share",
      "n_units": 3103,
      "weighting": "Vote-weighted",
      "year": 2020
    },
    {
      "axis": "Education",
      "band_high": 0.722,
      "band_low": 0.604,
      "conf_high": 0.722,
      "conf_low": 0.604,
      "estimate": 0.663,
      "first_high": null,
      "first_low": null,
      "first_share": 59.9,
      "measure": "College share",
      "n_units": 3103,
      "weighting": "Vote-weighted",
      "year": 2024
    },
    {
      "axis": "Income",
      "band_high": null,
      "band_low": null,
      "conf_high": 0.198,
      "conf_low": -0.005,
      "estimate": 0.096,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median household income",
      "n_units": 3113,
      "weighting": "Vote-weighted",
      "year": 2000
    },
    {
      "axis": "Income",
      "band_high": null,
      "band_low": null,
      "conf_high": 0.231,
      "conf_low": 0.01,
      "estimate": 0.121,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median household income",
      "n_units": 3113,
      "weighting": "Vote-weighted",
      "year": 2004
    },
    {
      "axis": "Income",
      "band_high": null,
      "band_low": null,
      "conf_high": 0.28,
      "conf_low": 0.05,
      "estimate": 0.165,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median household income",
      "n_units": 3113,
      "weighting": "Vote-weighted",
      "year": 2008
    },
    {
      "axis": "Income",
      "band_high": null,
      "band_low": null,
      "conf_high": 0.269,
      "conf_low": 0.044,
      "estimate": 0.157,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median household income",
      "n_units": 3112,
      "weighting": "Vote-weighted",
      "year": 2012
    },
    {
      "axis": "Income",
      "band_high": null,
      "band_low": null,
      "conf_high": 0.439,
      "conf_low": 0.194,
      "estimate": 0.316,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median household income",
      "n_units": 3111,
      "weighting": "Vote-weighted",
      "year": 2016
    },
    {
      "axis": "Income",
      "band_high": null,
      "band_low": null,
      "conf_high": 0.495,
      "conf_low": 0.29,
      "estimate": 0.392,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median household income",
      "n_units": 3103,
      "weighting": "Vote-weighted",
      "year": 2020
    },
    {
      "axis": "Income",
      "band_high": null,
      "band_low": null,
      "conf_high": 0.504,
      "conf_low": 0.29,
      "estimate": 0.397,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median household income",
      "n_units": 3103,
      "weighting": "Vote-weighted",
      "year": 2024
    },
    {
      "axis": "Race and ethnicity",
      "band_high": -0.357,
      "band_low": -0.681,
      "conf_high": -0.357,
      "conf_low": -0.681,
      "estimate": -0.519,
      "first_high": null,
      "first_low": null,
      "first_share": 86.9,
      "measure": "Non-Hispanic white share",
      "n_units": 3113,
      "weighting": "Vote-weighted",
      "year": 2000
    },
    {
      "axis": "Race and ethnicity",
      "band_high": -0.35,
      "band_low": -0.645,
      "conf_high": -0.35,
      "conf_low": -0.645,
      "estimate": -0.498,
      "first_high": null,
      "first_low": null,
      "first_share": 76.9,
      "measure": "Non-Hispanic white share",
      "n_units": 3113,
      "weighting": "Vote-weighted",
      "year": 2004
    },
    {
      "axis": "Race and ethnicity",
      "band_high": -0.418,
      "band_low": -0.671,
      "conf_high": -0.418,
      "conf_low": -0.671,
      "estimate": -0.545,
      "first_high": null,
      "first_low": null,
      "first_share": 75.8,
      "measure": "Non-Hispanic white share",
      "n_units": 3113,
      "weighting": "Vote-weighted",
      "year": 2008
    },
    {
      "axis": "Race and ethnicity",
      "band_high": -0.467,
      "band_low": -0.721,
      "conf_high": -0.467,
      "conf_low": -0.721,
      "estimate": -0.594,
      "first_high": null,
      "first_low": null,
      "first_share": 93.3,
      "measure": "Non-Hispanic white share",
      "n_units": 3112,
      "weighting": "Vote-weighted",
      "year": 2012
    },
    {
      "axis": "Race and ethnicity",
      "band_high": -0.573,
      "band_low": -0.791,
      "conf_high": -0.573,
      "conf_low": -0.791,
      "estimate": -0.682,
      "first_high": null,
      "first_low": null,
      "first_share": 78.7,
      "measure": "Non-Hispanic white share",
      "n_units": 3111,
      "weighting": "Vote-weighted",
      "year": 2016
    },
    {
      "axis": "Race and ethnicity",
      "band_high": -0.514,
      "band_low": -0.752,
      "conf_high": -0.514,
      "conf_low": -0.752,
      "estimate": -0.633,
      "first_high": null,
      "first_low": null,
      "first_share": 22.6,
      "measure": "Non-Hispanic white share",
      "n_units": 3103,
      "weighting": "Vote-weighted",
      "year": 2020
    },
    {
      "axis": "Race and ethnicity",
      "band_high": -0.44,
      "band_low": -0.707,
      "conf_high": -0.44,
      "conf_low": -0.707,
      "estimate": -0.574,
      "first_high": null,
      "first_low": null,
      "first_share": 2.2,
      "measure": "Non-Hispanic white share",
      "n_units": 3103,
      "weighting": "Vote-weighted",
      "year": 2024
    },
    {
      "axis": "Age",
      "band_high": null,
      "band_low": null,
      "conf_high": 0.036,
      "conf_low": -0.187,
      "estimate": -0.076,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median age",
      "n_units": 3113,
      "weighting": "Vote-weighted",
      "year": 2000
    },
    {
      "axis": "Age",
      "band_high": null,
      "band_low": null,
      "conf_high": -0.007,
      "conf_low": -0.219,
      "estimate": -0.113,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median age",
      "n_units": 3113,
      "weighting": "Vote-weighted",
      "year": 2004
    },
    {
      "axis": "Age",
      "band_high": null,
      "band_low": null,
      "conf_high": -0.117,
      "conf_low": -0.316,
      "estimate": -0.216,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median age",
      "n_units": 3113,
      "weighting": "Vote-weighted",
      "year": 2008
    },
    {
      "axis": "Age",
      "band_high": null,
      "band_low": null,
      "conf_high": -0.128,
      "conf_low": -0.351,
      "estimate": -0.24,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median age",
      "n_units": 3112,
      "weighting": "Vote-weighted",
      "year": 2012
    },
    {
      "axis": "Age",
      "band_high": null,
      "band_low": null,
      "conf_high": -0.255,
      "conf_low": -0.464,
      "estimate": -0.36,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median age",
      "n_units": 3111,
      "weighting": "Vote-weighted",
      "year": 2016
    },
    {
      "axis": "Age",
      "band_high": null,
      "band_low": null,
      "conf_high": -0.236,
      "conf_low": -0.435,
      "estimate": -0.335,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median age",
      "n_units": 3103,
      "weighting": "Vote-weighted",
      "year": 2020
    },
    {
      "axis": "Age",
      "band_high": null,
      "band_low": null,
      "conf_high": -0.214,
      "conf_low": -0.406,
      "estimate": -0.31,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median age",
      "n_units": 3103,
      "weighting": "Vote-weighted",
      "year": 2024
    },
    {
      "axis": "Rurality",
      "band_high": -0.357,
      "band_low": -0.565,
      "conf_high": -0.357,
      "conf_low": -0.565,
      "estimate": -0.461,
      "first_high": null,
      "first_low": null,
      "first_share": 13.2,
      "measure": "Rural share",
      "n_units": 3113,
      "weighting": "Vote-weighted",
      "year": 2000
    },
    {
      "axis": "Rurality",
      "band_high": -0.376,
      "band_low": -0.55,
      "conf_high": -0.376,
      "conf_low": -0.55,
      "estimate": -0.463,
      "first_high": null,
      "first_low": null,
      "first_share": 21.6,
      "measure": "Rural share",
      "n_units": 3113,
      "weighting": "Vote-weighted",
      "year": 2004
    },
    {
      "axis": "Rurality",
      "band_high": -0.429,
      "band_low": -0.603,
      "conf_high": -0.429,
      "conf_low": -0.603,
      "estimate": -0.516,
      "first_high": null,
      "first_low": null,
      "first_share": 23.6,
      "measure": "Rural share",
      "n_units": 3113,
      "weighting": "Vote-weighted",
      "year": 2008
    },
    {
      "axis": "Rurality",
      "band_high": -0.437,
      "band_low": -0.611,
      "conf_high": -0.437,
      "conf_low": -0.611,
      "estimate": -0.524,
      "first_high": null,
      "first_low": null,
      "first_share": 6.6,
      "measure": "Rural share",
      "n_units": 3112,
      "weighting": "Vote-weighted",
      "year": 2012
    },
    {
      "axis": "Rurality",
      "band_high": -0.557,
      "band_low": -0.743,
      "conf_high": -0.557,
      "conf_low": -0.743,
      "estimate": -0.65,
      "first_high": null,
      "first_low": null,
      "first_share": 20.3,
      "measure": "Rural share",
      "n_units": 3111,
      "weighting": "Vote-weighted",
      "year": 2016
    },
    {
      "axis": "Rurality",
      "band_high": -0.58,
      "band_low": -0.756,
      "conf_high": -0.58,
      "conf_low": -0.756,
      "estimate": -0.668,
      "first_high": null,
      "first_low": null,
      "first_share": 60.1,
      "measure": "Rural share",
      "n_units": 3103,
      "weighting": "Vote-weighted",
      "year": 2020
    },
    {
      "axis": "Rurality",
      "band_high": -0.564,
      "band_low": -0.741,
      "conf_high": -0.564,
      "conf_low": -0.741,
      "estimate": -0.653,
      "first_high": null,
      "first_low": null,
      "first_share": 37.9,
      "measure": "Rural share",
      "n_units": 3103,
      "weighting": "Vote-weighted",
      "year": 2024
    },
    {
      "axis": "Education",
      "band_high": 0.195,
      "band_low": 0.005,
      "conf_high": 0.195,
      "conf_low": 0.005,
      "estimate": 0.1,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "College share",
      "n_units": 3113,
      "weighting": "County-equal",
      "year": 2000
    },
    {
      "axis": "Education",
      "band_high": 0.294,
      "band_low": 0.127,
      "conf_high": 0.294,
      "conf_low": 0.127,
      "estimate": 0.211,
      "first_high": null,
      "first_low": null,
      "first_share": 1.6,
      "measure": "College share",
      "n_units": 3113,
      "weighting": "County-equal",
      "year": 2004
    },
    {
      "axis": "Education",
      "band_high": 0.374,
      "band_low": 0.245,
      "conf_high": 0.374,
      "conf_low": 0.245,
      "estimate": 0.31,
      "first_high": null,
      "first_low": null,
      "first_share": 11.6,
      "measure": "College share",
      "n_units": 3113,
      "weighting": "County-equal",
      "year": 2008
    },
    {
      "axis": "Education",
      "band_high": 0.378,
      "band_low": 0.232,
      "conf_high": 0.378,
      "conf_low": 0.232,
      "estimate": 0.305,
      "first_high": null,
      "first_low": null,
      "first_share": 1.1,
      "measure": "College share",
      "n_units": 3112,
      "weighting": "County-equal",
      "year": 2012
    },
    {
      "axis": "Education",
      "band_high": 0.546,
      "band_low": 0.378,
      "conf_high": 0.546,
      "conf_low": 0.378,
      "estimate": 0.462,
      "first_high": null,
      "first_low": null,
      "first_share": 5.5,
      "measure": "College share",
      "n_units": 3111,
      "weighting": "County-equal",
      "year": 2016
    },
    {
      "axis": "Education",
      "band_high": 0.611,
      "band_low": 0.448,
      "conf_high": 0.611,
      "conf_low": 0.448,
      "estimate": 0.53,
      "first_high": null,
      "first_low": null,
      "first_share": 33.8,
      "measure": "College share",
      "n_units": 3103,
      "weighting": "County-equal",
      "year": 2020
    },
    {
      "axis": "Education",
      "band_high": 0.638,
      "band_low": 0.474,
      "conf_high": 0.638,
      "conf_low": 0.474,
      "estimate": 0.556,
      "first_high": null,
      "first_low": null,
      "first_share": 58.8,
      "measure": "College share",
      "n_units": 3103,
      "weighting": "County-equal",
      "year": 2024
    },
    {
      "axis": "Income",
      "band_high": null,
      "band_low": null,
      "conf_high": 0.137,
      "conf_low": -0.101,
      "estimate": 0.018,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median household income",
      "n_units": 3113,
      "weighting": "County-equal",
      "year": 2000
    },
    {
      "axis": "Income",
      "band_high": null,
      "band_low": null,
      "conf_high": 0.141,
      "conf_low": -0.08,
      "estimate": 0.031,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median household income",
      "n_units": 3113,
      "weighting": "County-equal",
      "year": 2004
    },
    {
      "axis": "Income",
      "band_high": null,
      "band_low": null,
      "conf_high": 0.222,
      "conf_low": 0.01,
      "estimate": 0.116,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median household income",
      "n_units": 3113,
      "weighting": "County-equal",
      "year": 2008
    },
    {
      "axis": "Income",
      "band_high": null,
      "band_low": null,
      "conf_high": 0.21,
      "conf_low": -0.046,
      "estimate": 0.082,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median household income",
      "n_units": 3112,
      "weighting": "County-equal",
      "year": 2012
    },
    {
      "axis": "Income",
      "band_high": null,
      "band_low": null,
      "conf_high": 0.334,
      "conf_low": 0.048,
      "estimate": 0.191,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median household income",
      "n_units": 3111,
      "weighting": "County-equal",
      "year": 2016
    },
    {
      "axis": "Income",
      "band_high": null,
      "band_low": null,
      "conf_high": 0.411,
      "conf_low": 0.142,
      "estimate": 0.277,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median household income",
      "n_units": 3103,
      "weighting": "County-equal",
      "year": 2020
    },
    {
      "axis": "Income",
      "band_high": null,
      "band_low": null,
      "conf_high": 0.434,
      "conf_low": 0.173,
      "estimate": 0.304,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median household income",
      "n_units": 3103,
      "weighting": "County-equal",
      "year": 2024
    },
    {
      "axis": "Race and ethnicity",
      "band_high": -0.259,
      "band_low": -0.541,
      "conf_high": -0.259,
      "conf_low": -0.541,
      "estimate": -0.4,
      "first_high": null,
      "first_low": null,
      "first_share": 98.2,
      "measure": "Non-Hispanic white share",
      "n_units": 3113,
      "weighting": "County-equal",
      "year": 2000
    },
    {
      "axis": "Race and ethnicity",
      "band_high": -0.196,
      "band_low": -0.52,
      "conf_high": -0.196,
      "conf_low": -0.52,
      "estimate": -0.358,
      "first_high": null,
      "first_low": null,
      "first_share": 92.3,
      "measure": "Non-Hispanic white share",
      "n_units": 3113,
      "weighting": "County-equal",
      "year": 2004
    },
    {
      "axis": "Race and ethnicity",
      "band_high": -0.214,
      "band_low": -0.511,
      "conf_high": -0.214,
      "conf_low": -0.511,
      "estimate": -0.363,
      "first_high": null,
      "first_low": null,
      "first_share": 70,
      "measure": "Non-Hispanic white share",
      "n_units": 3113,
      "weighting": "County-equal",
      "year": 2008
    },
    {
      "axis": "Race and ethnicity",
      "band_high": -0.279,
      "band_low": -0.608,
      "conf_high": -0.279,
      "conf_low": -0.608,
      "estimate": -0.443,
      "first_high": null,
      "first_low": null,
      "first_share": 88.3,
      "measure": "Non-Hispanic white share",
      "n_units": 3112,
      "weighting": "County-equal",
      "year": 2012
    },
    {
      "axis": "Race and ethnicity",
      "band_high": -0.405,
      "band_low": -0.731,
      "conf_high": -0.405,
      "conf_low": -0.731,
      "estimate": -0.568,
      "first_high": null,
      "first_low": null,
      "first_share": 85.2,
      "measure": "Non-Hispanic white share",
      "n_units": 3111,
      "weighting": "County-equal",
      "year": 2016
    },
    {
      "axis": "Race and ethnicity",
      "band_high": -0.332,
      "band_low": -0.727,
      "conf_high": -0.332,
      "conf_low": -0.727,
      "estimate": -0.53,
      "first_high": null,
      "first_low": null,
      "first_share": 52,
      "measure": "Non-Hispanic white share",
      "n_units": 3103,
      "weighting": "County-equal",
      "year": 2020
    },
    {
      "axis": "Race and ethnicity",
      "band_high": -0.285,
      "band_low": -0.707,
      "conf_high": -0.285,
      "conf_low": -0.707,
      "estimate": -0.496,
      "first_high": null,
      "first_low": null,
      "first_share": 30,
      "measure": "Non-Hispanic white share",
      "n_units": 3103,
      "weighting": "County-equal",
      "year": 2024
    },
    {
      "axis": "Age",
      "band_high": null,
      "band_low": null,
      "conf_high": -0.054,
      "conf_low": -0.283,
      "estimate": -0.169,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median age",
      "n_units": 3113,
      "weighting": "County-equal",
      "year": 2000
    },
    {
      "axis": "Age",
      "band_high": null,
      "band_low": null,
      "conf_high": -0.031,
      "conf_low": -0.249,
      "estimate": -0.14,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median age",
      "n_units": 3113,
      "weighting": "County-equal",
      "year": 2004
    },
    {
      "axis": "Age",
      "band_high": null,
      "band_low": null,
      "conf_high": -0.071,
      "conf_low": -0.255,
      "estimate": -0.163,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median age",
      "n_units": 3113,
      "weighting": "County-equal",
      "year": 2008
    },
    {
      "axis": "Age",
      "band_high": null,
      "band_low": null,
      "conf_high": -0.084,
      "conf_low": -0.268,
      "estimate": -0.176,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median age",
      "n_units": 3112,
      "weighting": "County-equal",
      "year": 2012
    },
    {
      "axis": "Age",
      "band_high": null,
      "band_low": null,
      "conf_high": -0.183,
      "conf_low": -0.331,
      "estimate": -0.257,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median age",
      "n_units": 3111,
      "weighting": "County-equal",
      "year": 2016
    },
    {
      "axis": "Age",
      "band_high": null,
      "band_low": null,
      "conf_high": -0.149,
      "conf_low": -0.301,
      "estimate": -0.225,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median age",
      "n_units": 3103,
      "weighting": "County-equal",
      "year": 2020
    },
    {
      "axis": "Age",
      "band_high": null,
      "band_low": null,
      "conf_high": -0.118,
      "conf_low": -0.272,
      "estimate": -0.195,
      "first_high": null,
      "first_low": null,
      "first_share": 0,
      "measure": "Median age",
      "n_units": 3103,
      "weighting": "County-equal",
      "year": 2024
    },
    {
      "axis": "Rurality",
      "band_high": -0.18,
      "band_low": -0.357,
      "conf_high": -0.18,
      "conf_low": -0.357,
      "estimate": -0.269,
      "first_high": null,
      "first_low": null,
      "first_share": 1.8,
      "measure": "Rural share",
      "n_units": 3113,
      "weighting": "County-equal",
      "year": 2000
    },
    {
      "axis": "Rurality",
      "band_high": -0.179,
      "band_low": -0.336,
      "conf_high": -0.179,
      "conf_low": -0.336,
      "estimate": -0.257,
      "first_high": null,
      "first_low": null,
      "first_share": 6.1,
      "measure": "Rural share",
      "n_units": 3113,
      "weighting": "County-equal",
      "year": 2004
    },
    {
      "axis": "Rurality",
      "band_high": -0.26,
      "band_low": -0.395,
      "conf_high": -0.26,
      "conf_low": -0.395,
      "estimate": -0.328,
      "first_high": null,
      "first_low": null,
      "first_share": 18.5,
      "measure": "Rural share",
      "n_units": 3113,
      "weighting": "County-equal",
      "year": 2008
    },
    {
      "axis": "Rurality",
      "band_high": -0.286,
      "band_low": -0.424,
      "conf_high": -0.286,
      "conf_low": -0.424,
      "estimate": -0.355,
      "first_high": null,
      "first_low": null,
      "first_share": 10.6,
      "measure": "Rural share",
      "n_units": 3112,
      "weighting": "County-equal",
      "year": 2012
    },
    {
      "axis": "Rurality",
      "band_high": -0.417,
      "band_low": -0.555,
      "conf_high": -0.417,
      "conf_low": -0.555,
      "estimate": -0.486,
      "first_high": null,
      "first_low": null,
      "first_share": 9.2,
      "measure": "Rural share",
      "n_units": 3111,
      "weighting": "County-equal",
      "year": 2016
    },
    {
      "axis": "Rurality",
      "band_high": -0.449,
      "band_low": -0.586,
      "conf_high": -0.449,
      "conf_low": -0.586,
      "estimate": -0.518,
      "first_high": null,
      "first_low": null,
      "first_share": 14.2,
      "measure": "Rural share",
      "n_units": 3103,
      "weighting": "County-equal",
      "year": 2020
    },
    {
      "axis": "Rurality",
      "band_high": -0.462,
      "band_low": -0.592,
      "conf_high": -0.462,
      "conf_low": -0.592,
      "estimate": -0.527,
      "first_high": null,
      "first_low": null,
      "first_share": 11.2,
      "measure": "Rural share",
      "n_units": 3103,
      "weighting": "County-equal",
      "year": 2024
    }
  ],
  "id": "corr_by_year",
  "notes": [
    "Election returns are full counts of votes, so they carry no sampling error. The 95% intervals use errors clustered by state, so they show how much a number depends on which states are in the set. They do not include the sampling error of the county characteristics, which are census and American Community Survey estimates interpolated between years.",
    "Each county's measures are those of the election year, interpolated between census and survey years; see Methods."
  ],
  "source": "County presidential returns, 2000-2024 (Algara and Amlani for 2000 to 2020; MIT Election Data and Science Lab for 2024), full counts of votes; county characteristics from the decennial census and American Community Survey, interpolated to election years; estimates weighted by two-party votes or counting each county once, as stated",
  "spec": {
    "annotations": [
      {
        "type": "y-rule",
        "y": 0
      }
    ],
    "colors": {
      "Age": "cat6",
      "Education": "cat1",
      "Income": "cat3",
      "Race and ethnicity": "cat7",
      "Rurality": "cat2"
    },
    "hi": "band_high",
    "lo": "band_low",
    "series": "axis",
    "x": "year",
    "xDomain": [
      1999,
      2025
    ],
    "xFormat": {
      "digits": 0
    },
    "xLabel": "Presidential election",
    "xTicks": "data",
    "y": "estimate",
    "yDomain": [
      -0.8,
      0.8
    ],
    "yFormat": {
      "digits": 2,
      "signed": true
    },
    "yLabel": "Correlation with the county two-party margin (Democratic positive)"
  },
  "subtitle": "Correlation between a county characteristic and the county two-party margin in the same presidential election, 2000 to 2024, counties weighted by two-party votes",
  "table": {
    "columns": [
      {
        "format": {
          "digits": 0
        },
        "key": "year",
        "label": "Election"
      },
      {
        "key": "axis",
        "label": "Axis"
      },
      {
        "key": "measure",
        "label": "Measure"
      },
      {
        "key": "weighting",
        "label": "Weighting"
      },
      {
        "format": {
          "digits": 2,
          "signed": true
        },
        "key": "estimate",
        "label": "Correlation"
      },
      {
        "format": {
          "digits": 2,
          "signed": true
        },
        "key": "conf_low",
        "label": "Interval low"
      },
      {
        "format": {
          "digits": 2,
          "signed": true
        },
        "key": "conf_high",
        "label": "Interval high"
      },
      {
        "format": {
          "digits": 0
        },
        "key": "first_share",
        "label": "Share of resamples with the largest absolute correlation (%)"
      },
      {
        "format": {
          "thousands": true
        },
        "key": "n_units",
        "label": "Counties"
      }
    ]
  },
  "title": "Vote-weighted, the correlation of college share with the county margin rose from {fact:r_edu_2000_w} in 2000 to {fact:r_edu_2024_w} in 2024, while the white share's moved from {fact:r_race_2000_w} to {fact:r_race_2024_w}",
  "type": "line",
  "views": [
    {
      "filter": {
        "weighting": "Vote-weighted"
      },
      "id": "votes",
      "label": "Vote-weighted"
    },
    {
      "caption": "Counting each county once, the white share has the largest absolute correlation in the point estimates through 2016, college share and the white share are level in 2020, and college share is first in 2024, by {fact:gap_2024_u} over rural share, an interval that includes zero. Each line is the correlation across counties between one characteristic and the county two-party margin, with each county counted once; bands are 95% intervals with errors clustered by state, drawn for education, race and ethnicity and rurality only.",
      "filter": {
        "weighting": "County-equal"
      },
      "id": "equal",
      "label": "County-equal",
      "title": "Counting each county once, the correlation of college share with the county margin rose from {fact:r_edu_2000_u} in 2000 to {fact:r_edu_2024_u} in 2024"
    },
    {
      "caption": "The solid line is vote-weighted and the dashed line counts each county once; the gap between them is {fact:wd_edu_2000} in 2000 and {fact:wd_edu_2024} in 2024. Both lines plot the correlation across counties between the share of adults with a bachelor's degree and the county two-party margin; bands are 95% intervals with errors clustered by state.",
      "filter": {
        "axis": "Education"
      },
      "id": "education",
      "label": "Education, both weightings",
      "spec": {
        "colors": {
          "County-equal": "highlight",
          "Vote-weighted": "total"
        },
        "dashed": [
          "County-equal"
        ],
        "series": "weighting"
      },
      "title": "The correlation of college share with the county margin is lower in every election when each county counts once"
    },
    {
      "caption": "No axis leads in more than {fact:first_rural_2020_w} of resamples in 2020, and in 2024 college share leads in {fact:first_edu_2024_w} of resamples and rural share in {fact:first_rural_2024_w}. Each line is the share of {fact:n_reps} resamples of whole states in which the axis has the largest absolute correlation with the county margin, vote-weighted; the five lines sum to the whole in each election.",
      "filter": {
        "weighting": "Vote-weighted"
      },
      "id": "first_votes",
      "label": "Share of resamples in which an axis leads, vote-weighted",
      "spec": {
        "annotations": [],
        "hi": "first_high",
        "lo": "first_low",
        "y": "first_share",
        "yDomain": [
          0,
          100
        ],
        "yFormat": {
          "digits": 0,
          "suffix": "%"
        },
        "yLabel": "Share of resamples of states in which the axis has the largest absolute correlation (%)"
      },
      "title": "Vote-weighted, the white share has the largest absolute correlation in {fact:first_race_2016_w} of resamples in 2016, and college share has it in {fact:first_edu_2024_w} of resamples in 2024"
    },
    {
      "caption": "In 2020 the white share leads in {fact:first_race_2020_u} of resamples, college share in {fact:first_edu_2020_u} and rural share in {fact:first_rural_2020_u}. Each line is the share of {fact:n_reps} resamples of whole states in which the axis has the largest absolute correlation with the county margin, counting each county once.",
      "filter": {
        "weighting": "County-equal"
      },
      "id": "first_equal",
      "label": "Share of resamples in which an axis leads, county-equal",
      "spec": {
        "annotations": [],
        "hi": "first_high",
        "lo": "first_low",
        "y": "first_share",
        "yDomain": [
          0,
          100
        ],
        "yFormat": {
          "digits": 0,
          "suffix": "%"
        },
        "yLabel": "Share of resamples of states in which the axis has the largest absolute correlation (%)"
      },
      "title": "Counting each county once, the white share has the largest absolute correlation in {fact:first_race_2016_u} of resamples in 2016, and college share has it in {fact:first_edu_2024_u} of resamples in 2024"
    }
  ],
  "width": "wide"
}