======================================================================================================================= DOCUMENTATION FOR 1980-2022 GENERAL ELECTION TURNOUT RATES Version: 1.0 Date: 8/4/2023 Author: Michael McDonald, Professor of Political Science, University of Florida These data are released under an Creative Commons 4.0 Attribution License. See: https://creativecommons.org/licenses/by/4.0/ ======================================================================================================================= ============ VARIABLES ============ YEAR ======= Election year. STATE ======= Long state name. STATE_ABV ======= State two-letter postal abbreviation. TOTAL_BALLOTS_COUNTED ======= The total number of ballots counted by election officials in a November general election, either in official state reports or to the United States Election Assistance Commission (for decades 2000 and later). For 1990s and prior, I use reports generously provided by the Congressional Research Service and reported to the United States Congress. Notes: - Prior to 2010 fewer states report the total ballots counted in their official reports or to the United States Election Assistance Commission. I do not attempt to estimate missing data for states. I estimate a national number by computing the ratio of the VOTE_FOR_HIGHEST_OFFICE to the TOTAL_BALLOTS_CAST for reporting states and applying this ratio to the national sum of the VOTE_FOR_HIGHEST_OFFICE. To be transparent about this reporting I provide nulls for these states' data. - For more recent elections, observant researchers will notice that I include a "SOURCE" field in individual year datasets. I do not do include this field for this time-series. I primarily use SOURCE during the months after an election as a means to check for certification of election results so I can finalize the turnout rate numerator statistics and I provide them as a courtesy to interested researchers. - I make various adjustments to 2010-present election results. I recommend the documentation for these individual election years for more information. VOTE_FOR_HIGHEST_OFFICE ======= Vote for the highest statewide vote-getting office. In a presidential election this is almost always the presidential election and in a midterm it is US Senator or governor. Some states do not hold statewide governor elections in even years and may not have a US Senate election due to the staggered six-year terms. For these states, usually the combined totals for US Congress are used unless there is another statewide office which receives more votes. VAP ======= Voting-age population. Residential population age 18 and older. Source: Various July 1 Census Bureau population estimates. Census demographers take the last decennial census as a base and apply birth rates, death rates, and immigration rates drawn from various sources. The decennial census is not adjusted for under/over counts of various populations, a known issue with the Census. Most likely, the population estimates undercount the population of the United States. With each July 1 update to the population estimates, the Census Bureau not only releases a new estimate, but also updates prior years' population estimates spanning back to the last decennial census. The methodology is not transparent, but I imagine this process reflects updates to various primary birth, death, and immigration administrative record sources. When these data new become available, I typically update prior election years' voting-age population estimates. By the end of a decade, the population estimates become less reliable and do not match perfectly the population count revealed in the next decennial census. A few years following the decennial census, the Census Bureau releases the "intercensial adjustments" which adjusts the prior decade's population estimates to align with the bookend of decennial census population statistics. The release of these data typically mark the last update for my voting-age population estimates. NONCITIZEN_PCT ======= Percent non-citizen. Source: - For 2006 and after I use non-citizen VAP estimates from various 1-Year American Community Surveys. - Prior to 2006, I use interpolation between various decennial census citizenship estimates (and the 2005 American Community Survey for 2000, 2002, and 2004). The American Community Survey is a replacement for the now-defunct decennial census long-form 1 in 100 sample, which is where citizenship was previously asked. - In my prior published work, I used citizenship responses to the Current Population Survey's Voting and Registration Supplement, which includes a citizenship question. This is a valid citizenship estimate, but changes to the Census Bureau's methodology in the early 1990s lead me to believe that the interpolation method between decennial censuses yields a more reliable measure. - While outside the scope of this dataset, the 1960 decennial census did not include a citizenship question. This is a barrier to creating a more complete historical time-series for state citizen voting-age population estimates. - For presentation purposes, I only provide 3 signfiicant digits. The actual estimates reflect more precise citizenship percentage estimates. Further worthy of note is the Census Bureau weights the American Community Survey (and the Current Population Survey) to age-sex-race population estimates available at the time of the release of these surveys. The Census Bureau does not retroactively reweight these surveys to population estimate updates. It is for this reason I apply the noncitizen percentage to the VAP rather than directly using total citizen population estimates available from the American Community Survey. INELIGIBLE_PRISON ======= Ineligible felons in prison. A '0' or Null denotes a state either does not disenfranchise felony prisoners or has no felony prisoners (this is true for DC). Starting in 2010, the Bureau of Justice Statistics began reporting citizen prison populations and voting-age populations. I subtract these populations from the prison statistics. Source for prison statistics: Various year-end DOJ prison reports Source for state felon disenfranchisement laws: National Conference of State Legislatures. See: https://www.ncsl.org/elections-and-campaigns/felon-voting-rights Note: - State-level reports for prison populations do not exist prior to 1978. INELIGIBLE_PROBATION ======= Ineligbile felons who are on probation. A '0' denotes a state does not disenfranchise felons on probation. Source for probation statistics: Various year-end DOJ probation and parole reports. Source for state felon disenfranchisement laws: National Conference of State Legislatures. See: https://www.ncsl.org/elections-and-campaigns/felon-voting-rights The Department of Justice estimates that felons constitutue typically constitute between 50% and 60% of the total probation population. I apply this adjustment to various years' ineligible felon statistics. Note: - These reports do not exist prior to 1980, which is the primary reason I do not extend this time-series before 1980. INELIGIBLE_PAROLE ======= Ineligbile felons who are on parole. A '0' denotes a state does not disenfranchise felons on parole. Source for parole statistics: Various year-end DOJ probatiton and parole reports Source for state felon disenfranchisement laws: National Conference of State Legislatures. See: https://www.ncsl.org/elections-and-campaigns/felon-voting-rights Note: - These reports do not exist prior to 1980, which is the primary reason I do not extend this time-series before 1980. INELIGIBLE_FELONS_TOTAL ======= Total number of ineligible felons, the sum of INELIGIBLE_PRISON + INELIGIBLE_PROBATION + INELIGIBLE_PAROLE Notes: - I do not have a reliable method of estimating the number of felons who are subjected to some form of post-correctional disenfranchisement, so I choose not to make this adjustment. - I add the number of prisoners in the custody of the US government to the national count of ineligible felony prisoners. ELIGIBLE_OVERSEAS ======= Various sources, such as State Department reports of overseas citizens (typically permanent overseas residents who notify a country's embassy), Defense Department deployment reports, and in more recent years estimates from the Defense Department's Voting Assistance Commission, which is charged with facilitating voting for service members and overseas civilians. VEP ======= Voting-eligible population. For state VEP estimates use the following calculation: VAP*(1-NONCITIZEN_PCT) - INELIGIBLE_FELONS_TOTAL. For national VEP estimates use the following calculation: VAP*(1-NONCITIZEN_PCT) - INELIGIBLE_FELONS_TOTAL + OVERSEAS_ELIGIBLE VEP_TURNOUT_RATE ======= Turnout rate for those eligible to vote. Calulcated as TOTAL_BALLOTS_COUNTED/VEP. - Before 2010, an increasing number of states do not report TOTAL_BALLOTS_COUNTED and I do not attempt to estimate this number. For researchers who wish to analyze pre-2010 rates, I recommend either substituting the VOTE_FOR_HIGHEST_OFFICE for TOTAL_BALLOTS_COUNTED, or estimating TOTAL_BALLOTS_COUNTED from the ratio of TOTAL_BALLOTS_COUNTED to VOTE_FOR_HIGHEST_OFFICE from states that report both numbers. VAP_TURNOUT_RATE ======= Turnout rate for those of voting-age. Calulcated as TOTAL_BALLOTS_COUNTED/VAP Note: - The VEP turnout rate is the preferred turnout rate. I provide VAP turnout rates for some stakeholders, such as pollsters, who use VAP turnout rates.