United States Elections Project
   
 

Voter Turnout Frequently Asked Questions

What is the "Vote for Highest Office" and "Total Ballots Cast"?
What is the voting-age population or VAP?
What is the voting-eligible population or VEP?
Why should I care if turnout rates are calculated as a percentage of VAP or VEP?
How about calculating turnout rates for registered voters?
OK, I'm sold...How do you construct VEP? 
Where can I get your voter turnout data?
How is the Voting-Age Population constructed?
Why is the Current Population Survey VAP estimate lower than your estimate?
What happens to the VAP after the next census?
Where do you get your estimate of the non-citizen population?
How do you estimate the number of ineligible felons?
Do you correct for the number of permanently disfranchised felons?
How do you estimate the overseas voting-eligible population?
Why don't you make a state level adjustment for the overseas population?
What is left out of the voting-eligible population?
Where can I learn more?
What are some other useful data sources on voting?
Why didn't someone think of this before?
How do these turnout numbers differ from those produced by the Center for the Study of the American Electorate?
How do these turnout rates differ from those reported by the Current Population Survey?
I want county VEP estimates!

What is the "Vote for Highest Office" and "Total Ballots Cast"?

The Vote for Highest Office is the traditional reported number of people who voted in a given election. In presidential election years, the vote for highest office is simply the presidential vote. In a non-presidential election year the vote for highest office is the largest vote total for a statewide office such as governor or US Senator. When no statewide office is on the ballot, the sum of the congressional races is used instead.  In 2006, I changed this methodology slightly to use the sum of the congressional races if they exceeded a statewide office, as occurred with Indiana's uncompetitive US Senate race.

Some people do not cast a vote, even for president. Some failures to record votes are true errors, such as unrecorded votes originating from the infamous hanging chads of the 2000 Florida election. It is important to realize that some people intentionally abstain. for example, the 2004 presidential election 3,688 Nevadans voted for "None of These Candidates" (Nevada is the only state that allows this option). Under-Votes are such blank or indecernable votes. Over-votes occur when a voter selects multiple candidates when only one is acceptable. 

A better measure of participation might be Total Ballots Cast, which includes all under-votes and over-votes, but does not include rejected provisional ballots, which are often cast when a voter's eligibility is in question. (Residual vote is the difference between total ballots cast and vote for highest office.) The problem is that most, but not all, states report the total ballots cast.

The good news is that an increasing number of states report total ballots cast, which makes it possible to estimate the total ballots cast for the few non-reporting states. National total ballots cast is therefore an estimate using the correspondence between the vote for highest office and total turnout for the states that provide both numbers. 

What is the voting-age population or VAP?

The voting-age population is defined by the Bureau of the Census as everyone residing in the United States, age 18 and older. Before 1971, the voting-age population was age 21 and older for most states.

What is the voting-eligible population or VEP?

The voting-eligible population is a phrase I coined to describe the population that is eligible to vote. Counted among the voting-age population are persons who are ineligible to vote, such as non-citizens, felons (depending on state law), and mentally incapacitated persons. Not counted are persons in the military or civilians living overseas.

The voting-age population is appropriately adjusted in order to arrive at the voting-eligible population, as described below. For maximum transparency, the voting-age population estimates and statistics used to modify it to arrive at the voting-eligible population are provided in each year's table of turnout rates.

Why should I care if turnout rates are calculated as a percentage of VAP or VEP?

The most valid turnout rates over time and across states are calculated using voting-eligible population.

Declining turnout rates, post-1971, are entirely explained by the increase in the ineligible population.  In 1972, the non-citizen population of the United States was less than 2 percent of VAP and in 2004 it was nearly 8.5 percent of VAP.  The percent of non-felons among the VAP have increased from .5 to about 1 percent of the VAP since the mid-1980s.

State turnout rates are not comparable using VAP since the ineligible population is not uniformly distributed across the United States.  For example, nearly 20 percent of California's voting-age population is ineligible to vote because they are not citizens.

How about calculating turnout rates for registered voters?

Many states do this, but while it is a useful number from an administrative standpoint, it is practically useless from a policy standpoint. Voter registration rolls contain 'deadwood' - people who are registered at an address but no longer reside there, for whatever reason.  States vary on how well they maintain their registration rolls and have changed their purging procedures over time.  Thus, comparing state turnout rates based on voter registration is not informative. 

See a 2004 voter registration report for more information. (I plan to update this report for the 2008 election).

OK, I'm sold...How do you construct VEP? 

The voting-eligible population is constructed by adjusting the voting-age population for non-citizens and ineligible felons, depending on state law. National estimates are further adjusted for overseas eligible voters, but no state level adjustments are made since their is no reliable method of apportioning overseas voters to states. There are other adjustments that I am unable to made due to lack of available data, such as the number of permanently disfranchised felons, depending on state law, the number of persons who have been judged mentally incompetent by state law, and the number of persons who have moved within a state after the close of voter registration, depending on state law.

Note, if you have moved between states after a registration deadline federal law guarantees you the right to vote for president in your former state (see 42 USC § 1973aa-1(e)), many states have provisions that permit voting if you have moved within your state and have not updated your registration. Please check with your local election administrators for more information for your state's laws.

Where can I get your voter turnout data?

Tables for recent elections appear on this website.  A spreadsheet of national and state data, 1980-2006 are available in an Excel spreadsheet: Turnout 1980-2006.xls

How is the Voting-Age Population constructed?

The voting-age population estimates are generated from the July 1 Census Bureau population estimates by age-sex-race. Estimates for an election month are constructed by linearly interpolating the change between the most approximate pair of estimates. For example, to estimate the Nov 2004 VAP population, add to the July 1, 2004 estimate the following: the July 1, 2004 estimate minus the July 1, 2003 estimate, divide by 12, and then multiply by 4.  When forecasting, I adjust the most recent estimate in a similar manner.  

It should be emphasized that although these estimates are based on official Census Bureau reports, listed below, but they are not directly produced by the Census Bureau and should not be cited as a Census Bureau product.

  • Post-2000 population estimates are available from the Census Bureau's web site <LINK>.
  • The Census Bureau generated new population estimates from 1990-2000 based on the 2000 census (so-called "intercenisal adjustments") for NCHS <LINK>.

Why is the Current Population Survey VAP estimate lower than your estimate?

The Current Population Survey is a Census Bureau survey of the civilian non-institutional population of the United States. The VAP estimates are of the resident population of the United States, which includes military and other persons living in group quarters that are not in the Current Population Survey sample frame.

What happens to the VAP after the next census?

Current VAP population estimates are based on estimates of population change closely followed by the population estimates division at the Census Bureau. When a new census is conducted at the end of the decade, these population estimates are inevitably out of line with the actual census. The Census Bureau adjusts the previous decade's population estimates to conform with the new information revealed by the census.  This is called the intercensal adjustment. Thus, we will not have a definitive estimate of the turnout rate in the 2004 election until the intercensal adjustment is released sometime after 2010.  In the meantime, the population estimates serve as the best estimate of the voting-age population of the United States. 

Where do you get your estimate of the non-citizen population?

For 2000 and years prior, I linearly interpolate the non-citizen population among the VAP drawn from the census.  Following 2000, I use the Current Population Survey estimate of the non-citizen population.  

  • CPS surveys are available to ICPSR subscribers <LINK> and you may also download the full surveys from the Census Bureau's Data Ferrett system <LINK>.  Tables and other infromation are also available on the Census Bureau website <LINK>.
  • April, 1 2000 STF 4 file estimate of noncitizens, available through the Census Bureau "American FactFinder" <LINK>.

I construct a the percentage of non-citizens among the VAP by dividing the total estimated number of CPS non-citizens by the best VAP estimate. The percentages in the tables provided are thus somewhat misleading and seemingly inconsistent with the CPS, which simply reports the percentage of the CPS sample frame (this is an unfortunate legacy from reporting of the non-citizen numbers prior to 2000). 

I take this approach so because the CPS's sample frame is the non-institutional civilian population of the United States whereas the voting-age population estimate is for the resident population of the United States. Non-citizens can be found in nursing home, university dorms, and military quarters, but I suspect that the proportion of non-citizens is less than the non-institutional population. Thus, I am offering a conservative estimate of the non-citizen population with this method.  Furthermore, I subtract felon prisoners through a different method descibed below, so not counting institutionalized prisoners avoids double counting the small percentage of non-citizen prisoners.

How do you estimate the number of ineligible felons?

The number of ineligible felons depends on state law. Some state permanently disfranchise felons and a few let even prisoners vote. Statistics drawn from various Department of Justice reports which detail the prison, probation, and parole population of the United States are matched with these state laws to estimate the number of ineligible felons. 

The Department of Justice Bureau, Office of Justice Statistics releases numbers from Jan. 1 through Dec. 31 for a given year. I use the most approximate Jan. 1 values, where available, since these include revised Dec. 31 data from the previous year. I assume all Prisoners and Parolees are felons, and half of Probationers are felons (this estimate is drawn from DOJ reports).  A blank or zero indicates the category of felons is allowed to vote in that state. For the United States totals, I include persons in the Federal corrections system.  

Do you correct for the number of permanently disfranchised felons?

No, I do not estimate the number of "permanently" disfranchised felons in the 14 states that have some form of post-correctional voting restriction, since statistics on recidivism, deaths, and migration of felons are largely unknown.

For recent elections, Chris Uggen and Jeff Manza estimate the number of permanently disfranchised felons in the 2004 electon. Because their methodology has not been updated for a longer period of time, I have not adopted this adjustment.

How do you estimate the overseas voting-eligible population?

The overseas eligible population is drawn from various sources: Department of State Consular Services estimates of civil population abroad, DoD military deployment records, and Department of State foreign service employees. Unfortunately, Consular Services no longer published estimates of the private citizen overseas population. The Census Bureau provided a thorough overview of the problems associated with estimating overseas VAP <LINK> which concluded conducting a census of overseas citizens in impossible at this time.

For 2008, I have obtained estimates of overseas numbers from the Federal Voting Assistance Program, the federal entity created by the Uniform and Civilian Overseas Absentee Voting Act. These estimates are reported in letters FVAP sends to the states. These letters report internal statistics from the Department of Defense that list all overseas military personnel who claim a state as their residence. My understanding is that the Department of Defense estimates that three civilian dependents are stationed with every four deployed personnel. FVAP then apportions the number of civilians reported by the Department of State's Consular Services among the states using a formula based on the number of House seats within a state or territory. Note that FVAP, like the numbers reported here, use Consular Services reports that are now out of date.

I modify the FVAP numbers for my calculations. I use the number of deployed military personnel as reported by the Department of Defense. However, I deflate the civilian numbers by 25%, which is the proportion of minors among the United States resident citizen population. I have verified using statistics generated from overseas voters who have registered with the Overseas Vote Foundation that the proportion of overseas registrants claiming a state as their home is roughly equal to the proportion of a states resident citizen population, i.e., more people register in states with large populations. Instead of allocating overseas civlians based on congressional representation, I use a more direct proportion of a state's citizen population.

Why don't you make a state level adjustment for the overseas population?

To my knowledge, there is no method to apportion overseas VEP to the states. However, I am investigating methods to do so based on surveys of overseas citizens and tax returns. Preliminary evidence suggests that apportioning overseas voters roughly equal to the population of the states provides a reasonable estimate of the overseas citizens who claim a state as their residence. 

What is left out of the voting-eligible population?

  • I do not correct for the number of persons found mentally incapacitated by a court of law since there is no reliable estimate.  My guess from examination of National Health Statistics reports is that this population is somewhere around 250,000.
  • I do not correct for the number of unregisted and registered voters who - depending on states law - have moved between jurisdictions within a state who may not be able to vote because they moved after their state's registration deadline, or for midterm elections, persons who move between states after their new state's registration deadline. (Note, if this situation pertains to you, please check with your local election officials as many states have provisions that permit recent movers to vote.)
  • I do not correct number of permanently disfranchised felons.
  • I do not correct for the census net undercount of the population, since no state-level estimates are available at this time. 

Where can I learn more?

Primary research published by Dr. Michael P. McDonald at George Mason University (email:mmcdon@gmu.edu office phone: 703-993-4191) appears in the following journals:

  • Michael P. McDonald.  2002.  "The Turnout Rate Among Eligible Voters for U.S. States, 1980-2000."  State Politics and Policy Quarterly 2(2): 199-212.
  • Michael P. McDonald and Samuel Popkin. 2001. "The Myth of the Vanishing Voter." American Political Science Review 95(4): 963-974.

Secondary research on this topic appears in the following publications:

  • Michael P. McDonald.  (Forthcoming)  "Portable Voter Registration."  Political Behavior.
  • Michael P. McDonald.  (Forthcoming).  "American Voter Turnout in Historical Perspective."  In Oxford Handbook of American Elections and Political Behavior, Jan Leighley, ed.  Cambridge, UK: Oxford University Press.
  • Michael P. McDonald.  2007.  "The True Electorate: A Cross-Validation of Voter File and Election Poll Demographics."  Public Opinion Quarterly 71(4): 588-602.
  • Michael P. McDonald.  2006.  "Rocking the House: Competition and Turnout in the 2006 Midterm Election."  The Forum 4(3).
  • Michael P. McDonald.  2005.  "Reporting Bias" in Polling in America: An Encyclopedia of Public Opinion.  Benjamin Radcliff and Samuel Best, eds.  Westport, CT: Greenwood Press.
  • Michael P. McDonald. 2004. "Up, Up, and Away! Turnout in the 2004 Presidential Election." The Forum (2):4. Dec 2004.
  • Michael P. McDonald. 2003. "On the Over-Report Bias of the National Election Study." Political Analysis 11(2): 180-186.

What are some other useful data sources on voting?

Why didn't someone think of this before?

Walter Dean Burnham made adjustments for the non-citizen population until 1986. Those who study voter turnout are deeply indebted to his painstaking research to construct historical voter turnout rates. Unfortunately, he stopped writing on this topic in the mid-1980s and therefore missed the up-tick in the non-citizen and ineligible felon populations, and thus missed explaining the often-discussed decline in contemporary turnout rates. Recently, he has published updated turnout numbers in the Journal of the Historical Society. I am certain that if he had continued with his calculations after the 1990 census, he would have likely realized the ineligible population was growing at such a high rate it was artificially depressing the VAP turnout rate.

How do these turnout numbers differ from those produced by the Center for the Study of the American Electorate?

The Center for the Study of the American Electorate also produces national and state level turnout rates for the United States. There are a number of differences between these statistics:

  1. The numbers reported here are published in peer-reviewed journals, while those reported by the CSAE are not. The turnout rates here are reported on election night to all major media outlets through Edison Media Research, which conducts the national exit poll for the consortium of media organizations. (Disclaimer: Dr. McDonald consults for Edison Media Research on election night.) They have also been widely republished in textbooks and reference books and are used by various public interest groups.
  2. The turnout rates here adjust for non-citizens, ineligible felons, and overseas citizens. The CSAE only adjusts for non-citizens, and only began doing so after the data here were publicly released. CSAE asserts without proof that the ineligible felon and overseas citizen population "are roughly equal" (see p.2 of the 2006 report). However, even if this assertion is true nationally, it is almost certainly false within states as there is considerable variation among states as to who they disfranchise among their prisioners, parolees, and probationers.
  3. All primary sources used here to construct data are clearly identified and hyperlinks are provided where possible so that interested persons can replicate the analysis. Data are further provided in an easily assessible Excel spreadsheet. The CSAE does not have similar transparency.
  4. There are other technical details that are different. Reported here are the most recently available census population estimates and estimates of the non-citizen population while the CSAE projects forward the difference between the last two censuses. Thus, CSAE estimates would be reduced in accuracy if, say, immigration into the United States slowed due to an economic downturn or changes in U.S. immigration policy following 9/11.

How do these turnout rates differ from those reported by the Current Population Survey?

Another source for turnout and voter registration rates is the Current Population Survey's biannual reports on voting and registration. Indeed, these data are essentially the best information for turnout and registration rates among various demographic subgroups, such as the young or by race, which cannot be computed from aggregate data presented here. However, the CPS is a survey and subject to a phenomenon known as vote over-report bias, whereby a greater percentage of people report voting than the aggregate statistics reported here indicates. There are many explanations for this phenomenon, from lying by survey respondents to participation in surveys by people who are more participatory in other aspects, such as voting. The CPS has lower over-report bias than other surveys, such as the American National Election Study, but it is present on the CPS. The chart below illustrates the differences between the turnout rates reported here, and those reported by the ANES and CPS surveys.

After the data reported here were made public, the Census Bureau began reporting turnout rates for the citizen-voting age population, whereas previously they reported voting-age population turnout rates.

Over-report bias chart
This chart is adapted from Michael P. McDonald. 2003. "On the Over-Report Bias of the National Election Study." Political Analysis 11(2): 180-186 and Michael P. McDonald.  2005.  "Reporting Bias" in Polling in America: An Encyclopedia of Public Opinion.  Benjamin Radcliff and Samuel Best, eds.  Westport, CT: Greenwood Press.

Why doesn't the government produce these numbers?

That would save me a lot of time! Some states produce estimates of their eligible voters, but often do not explain how they arrived at their estimates. 

I want county VEP estimates!

So do I! The best estimate that can be constructed at the county level is citizen-voting age population.  Data available at National Historical Geographic Information System may be extremely helpful in constructing these data, and many other variables of interest.