{
  "alt": "Dumbbell chart with five rows for each of 2000 and 2024, giving the county-equal and vote-weighted correlations of each axis with the county margin. College share is higher vote-weighted in both years, {fact:r_edu_2000_w} against {fact:r_edu_2000_u} in 2000 and {fact:r_edu_2024_w} against {fact:r_edu_2024_u} in 2024, and rural share is more negative vote-weighted, {fact:r_rural_2000_w} against {fact:r_rural_2000_u} in 2000.",
  "caption": "For college share and rural share the vote-weighted and county-equal correlations differ more in 2000 than in 2024 ({fact:wd_rural_2000} and {fact:wd_rural_2024} for rural share), and for income and median age the differences are larger in absolute size in 2024 ({fact:wd_inc_2000} to {fact:wd_inc_2024} and {fact:wd_age_2000} to {fact:wd_age_2024}), the age difference with a changed sign. Each row plots two correlations across the same counties with 95% intervals with errors clustered by state, and the last column is their difference.",
  "data": [
    {
      "axis": "Education",
      "ce_high": 0.195,
      "ce_low": 0.005,
      "county_equal": 0.1,
      "election": "2000",
      "n_units": 3113,
      "vote_weighted": 0.282,
      "vw_high": 0.381,
      "vw_low": 0.182
    },
    {
      "axis": "Income",
      "ce_high": 0.137,
      "ce_low": -0.101,
      "county_equal": 0.018,
      "election": "2000",
      "n_units": 3113,
      "vote_weighted": 0.096,
      "vw_high": 0.198,
      "vw_low": -0.005
    },
    {
      "axis": "Race and ethnicity",
      "ce_high": -0.259,
      "ce_low": -0.541,
      "county_equal": -0.4,
      "election": "2000",
      "n_units": 3113,
      "vote_weighted": -0.519,
      "vw_high": -0.357,
      "vw_low": -0.681
    },
    {
      "axis": "Age",
      "ce_high": -0.054,
      "ce_low": -0.283,
      "county_equal": -0.169,
      "election": "2000",
      "n_units": 3113,
      "vote_weighted": -0.076,
      "vw_high": 0.036,
      "vw_low": -0.187
    },
    {
      "axis": "Rurality",
      "ce_high": -0.18,
      "ce_low": -0.357,
      "county_equal": -0.269,
      "election": "2000",
      "n_units": 3113,
      "vote_weighted": -0.461,
      "vw_high": -0.357,
      "vw_low": -0.565
    },
    {
      "axis": "Education",
      "ce_high": 0.638,
      "ce_low": 0.474,
      "county_equal": 0.556,
      "election": "2024",
      "n_units": 3103,
      "vote_weighted": 0.663,
      "vw_high": 0.722,
      "vw_low": 0.604
    },
    {
      "axis": "Income",
      "ce_high": 0.434,
      "ce_low": 0.173,
      "county_equal": 0.304,
      "election": "2024",
      "n_units": 3103,
      "vote_weighted": 0.397,
      "vw_high": 0.504,
      "vw_low": 0.29
    },
    {
      "axis": "Race and ethnicity",
      "ce_high": -0.285,
      "ce_low": -0.707,
      "county_equal": -0.496,
      "election": "2024",
      "n_units": 3103,
      "vote_weighted": -0.574,
      "vw_high": -0.44,
      "vw_low": -0.707
    },
    {
      "axis": "Age",
      "ce_high": -0.118,
      "ce_low": -0.272,
      "county_equal": -0.195,
      "election": "2024",
      "n_units": 3103,
      "vote_weighted": -0.31,
      "vw_high": -0.214,
      "vw_low": -0.406
    },
    {
      "axis": "Rurality",
      "ce_high": -0.462,
      "ce_low": -0.592,
      "county_equal": -0.527,
      "election": "2024",
      "n_units": 3103,
      "vote_weighted": -0.653,
      "vw_high": -0.564,
      "vw_low": -0.741
    }
  ],
  "id": "weighting_gap",
  "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.",
    "The difference in the last column is its own estimate; its 95% interval from a bootstrap over states is in the text."
  ],
  "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": {
    "deltaLabel": "Vote-weighted minus county-equal",
    "group": "election",
    "hi1": "ce_high",
    "hi2": "vw_high",
    "label1": "County-equal",
    "label2": "Vote-weighted",
    "lo1": "ce_low",
    "lo2": "vw_low",
    "n1": "n_units",
    "n2": "n_units",
    "reference": 0,
    "sort": "data",
    "x1": "county_equal",
    "x2": "vote_weighted",
    "xDomain": [
      -0.8,
      0.8
    ],
    "xFormat": {
      "digits": 2,
      "signed": true
    },
    "xLabel": "Correlation with the county two-party margin (Democratic positive)",
    "y": "axis"
  },
  "subtitle": "Correlation of each axis with the county two-party margin when counties are weighted by two-party votes and when each county counts once, 2000 and 2024",
  "title": "The vote-weighted correlation of college share exceeds the county-equal one by {fact:wd_edu_2000} in 2000 and by {fact:wd_edu_2024} in 2024",
  "type": "dumbbell",
  "width": "wide"
}