Overseas Citizen Population Analysis Report
This report is a detailed analysis of American voters overseas that estimates the population of eligible voters and includes results of the first-ever representative survey of registered overseas voters who requested an absentee ballot.
Frequently Asked Questions (Updated 10/18/16):
Q1) What were the primary objectives of the Overseas Citizen Population Analysis (OCPA) project?
A1) The Federal Voting Assistance Program (FVAP) conducted the 2014 OCPA to gain a greater understanding of the voting behavior of U.S. citizens living overseas during the 2014 General Election and fulfill a longstanding requirement for reporting registration and participation rates among overseas citizens pursuant to the Uniformed and Overseas Citizens Absentee Voting Act (UOCAVA). This research effort involved creating estimates of the number of U.S. citizens in each foreign country and conducting a survey of registered overseas voters. Results will allow FVAP to keep report of overseas voter participation rates over time and to determine what types of obstacles prevent overseas citizens from voting when they have a desire to do so.
Q2) Has anyone ever calculated the voting rate of overseas U.S. citizens before?
A2) There have been attempts to calculate an overall voting rate for overseas citizens; however, the 2014 OCPA was the first project that assembled the data necessary to estimate the overseas voting population, a critical requirement for moving forward. The subsequent effort of the 2014 OCPA was an actual survey of known overseas voters. Throughout this effort, FVAP and its researchers completed several subgoals, including:
- Estimating an aggregate voting rate of the overseas civilian population in the 2014 General Election. To calculate this, we divided the number of 2014 General Election ballots that were recorded by local election officials by our estimate of the total number of voting-eligible overseas citizens.
- Estimating the voting rate of the subset of voting-eligible overseas citizens who actively requested a ballot. To do this, we divided the number of 2014 General Election ballots that were requested by overseas citizens by the number of 2014 General Election ballots that were received from overseas citizens and recorded by local election officials.
- Estimating the voting rate among overseas citizens who were likely to have been interested in voting in the 2014 General Election.
Previous estimates of the overall voting rate have been produced by the Overseas Vote Foundation (OVF), which calculated an overseas voting rate of 6.8 percent in 2008.[i] This figure is fairly consistent with our estimated overseas voting rate of 4 percent in 2014, since presidential elections typically have a higher turnout than midterm elections.
These are general points of validation for our research including those that could draw an estimate of participation using available data on the number of returned ballots from overseas citizens reported to the United States Election Assistance Commission.
Q3) Has anyone else ever produced estimates of the number of eligible voters living overseas?
A3) The United Nations and World Bank have produced estimates of the number of U.S. born or U.S. citizens living overseas by country. FVAP produced estimates of the number of overseas U.S. citizens in 2013, but these estimates included all overseas U.S. citizens--not just those who were eligible to vote (i.e., those aged 18 or older). The U.S. Census Bureau piloted a program to assess the feasibility of including overseas citizens in its decennial census, but the study found that full implementation would be prohibitively expensive and would not produce a level of data quality consistent with Census standards.[ii] In previous years, the Department of State has released estimates of the number of overseas civilians; in 2011, this estimate was 6.3 million.[iii]
Q4) How are the current OCPA estimates of overseas citizens different from previous estimates?
A4) Previous estimates: (1) excluded key countries; (2) were produced for years before 2014; or (3) used data sources that had different definitions of U.S. citizens--for instance, some counted all U.S. citizens, some counted individuals who were born in the U.S., and some counted only U.S. citizens who did not have dual citizenship in their country of residence. In the 2014 OCPA, these issues were addressed by using a statistical tool. We assumed that the number of Americans who were counted in a specific country was directly related to:
- Whether the country counted U.S. citizens or U.S. born individuals;
- How the count was conducted, such as by a complete population census or through government registries;
- U.S. government records of overseas Americans in the country (e.g., foreign income tax filers, social security beneficiaries, civilian government employees, students); and
- A set of country characteristics that may affect immigration to or from that country, such as the distance from the United States, the quality of governance/institutions, and the primary language of the country.
We used all of these factors to estimate the number of U.S. citizens living in each foreign country, and we used a similar modeling approach to estimate the fraction of the total U.S. citizen population that was eligible to vote (i.e., aged 18 or older). Summing the voting-eligible population across countries provided an estimate of the total number of eligible overseas voters.
Q5) Has anyone ever conducted a survey of overseas U.S. citizens?
A5) The 2014 Overseas Citizen Population Survey (OCPS), a component of the larger OCPA project, was the first probability-based survey of overseas voters who requested an absentee ballot. It builds on past surveys conducted by FVAP and other organizations by using a more rigorous and inclusive sampling method. Specifically, in 2004, the Census Bureau conducted a pilot survey of overseas citizens in three countries, including overseas voters. This effort was not recommended for full implementation, given the difficulties in contacting overseas citizens as well as the cost and complexity of implementation. In the absence of a comprehensive list of all overseas citizens, the Census Bureau relied on third-party organizations and paid advertising to distribute questionnaires and to draw attention to the study. More recent efforts include the work of private organizations like the Overseas Vote Foundation (OVF), which conducts a voting survey of its website users, many of whom are current overseas voters. FVAP also conducted a survey of overseas citizens in 2004, which was distributed through embassies in various locations.[iv] While all of these studies provided valuable information, none met the "gold standard" criterion: selecting respondents from a list of people in a way that each has an equal chance, or probability, of being surveyed.
The primary difference between the 2014 OCPS and prior efforts is our objective to create a comprehensive list of overseas voters who requested an absentee ballot, and consequently, the population to which survey results can be generalized. The 2014 OCPS sample frame was developed by obtaining data from absentee ballot request files maintained by State and local election officials. A stratified sample of ballot requesters was then selected and invited to participate in the survey. Finally, the data were weighted to accurately reflect the absentee-ballot-requesting population as a whole and not just the survey respondents.
Q6) How do you know how many people voted from each country? Aren't voting records private?
A6) Estimates of the total number of votes that were counted from overseas civilians can be obtained from the U.S. Election Assistance Commission's (EAC) Election Administration and Voting Survey (EAVS), which is a survey of State and local election officials. However, this survey only reports the State of residence and not the country from which votes were submitted. Consequently, FVAP produced its own count of registered overseas voters by obtaining data from absentee ballot request files maintained by State and local election officials. Knowing how many voters were in each country allowed us to calculate country-specific voting rates.
Q7) Is the fraction of the overseas population that is of voting age really that small?
A7) First of all, keep in mind that the estimates of the total overseas citizen population and the eligible overseas population are just that--estimates. They are produced using sophisticated technical models, but ultimately, there will always be a level of uncertainty about the true number of overseas citizens. Our research team was surprised by the proportion of the overseas citizen population that was under the age of 18, but we think that it makes sense that the overseas citizen population would tend to be younger than the domestic population.
Children are considered to be U.S. citizens if at least one of their parents is a U.S. citizen or if they are born in the United States (regardless of whether their parents are U.S. citizens). This means that U.S. citizens who are married to a citizen of another country might have children who are eligible for U.S. citizenship, even though only one parent is a U.S. citizen. Similarly, parents who are citizens of another country could live in the United States for a short time, have children in the United States, and then migrate back to their home country. Their children would be U.S. citizens, even though the parents are not. These cases tend to skew the ratio of overseas U.S. citizens to be much younger than the domestic population in which both parents and children are typically U.S. citizens.
The OCPS collected data on the citizenship and family status of overseas citizens who requested an absentee ballot from an address outside of the United States. Among respondents who reported being married and having a child who was a U.S. citizen, approximately half (51.7 percent) indicated that their spouse does not have U.S. citizenship. While this sample of overseas ballot requesters is not necessarily representative of the general overseas population, there is little reason to think this proportion represents an overestimate relative to the general overseas population.
There is quite a bit of geographic variation in the proportion of overseas citizens who are aged 18 or older. For instance, 70 percent of U.S. citizens residing in Canada and 83 percent of those residing in Germany are estimated to be 18 or older, whereas only 7 percent of U.S. citizens residing in Mexico and 5 percent of those residing in India are estimated to be 18 or older. To some extent, this mirrors overall population trends in these countries, where relatively high birthrates in Mexico and India mean that the overall population tends to be younger in those countries. These statistics are also influenced by the return migration of children who are U.S. citizens born to parents who are not. In the case of Mexico, high levels of return migration have been specifically documented by PEW.[v]
Q8) That sounds like a very low voting rate. Are the obstacles to voting outside the United States really that large?
A8) We agree that this estimate is low compared to the voting rates estimated for other populations, including Active Duty Military (ADM). There are a multitude of factors that might be involved in producing this lower rate of participation amongst this overseas citizen population - obstacles in the process, lack of engagement, interest in midterm elections, etc.[vi] Now that we have a baseline to work with, we can assess the utility of those efforts to move the needle on this rate.
Q9) Where do these numbers come from? Doesn't the U.S. Department of State know how many Americans are abroad?
A9) The United States does not issue exit visas or monitor its citizens when they travel outside of the country. Individuals who travel or live abroad can register with their local embassy to receive alerts from the Department of State, but no official registry is kept. These registries include citizens on short-term travel as well as long-term relocation, and there is no process to remove duplicates if individuals register at more than one embassy or to remove people from the list who have returned to the United States. As noted in A3, the Department of State has released these estimates publicly in previous years but has recently discontinued that practice.
Because no official count of overseas citizens exists, our research team compiled data from many different sources to create a statistical estimate of the number of overseas citizens. These sources include registries and censuses kept by foreign governments, administrative records from U.S. agencies, and economic, social, and geographic variables for each foreign country.
Q10) Why aren't military Service members included? How did you make sure they were excluded? Where can I find info about (overseas) military voting?
A10) FVAP's mission is to provide voting assistance to military Service members, their families, and overseas citizens. This particular research project focused on the last group as FVAP currently reports registration and participation rates for the military in its post-election Congressional reporting. To control for any overlap, the 2014 OCPS asks a question regarding why the respondent was living overseas, and respondents who answered that they were in the military or were the spouse of a military Service member were excluded from the analysis. More information about military voting can be found online at FVAP.gov.
Q11) What is the difference between an overseas citizen and an overseas voter? Can all overseas citizens also be considered overseas voters?
A11) An overseas citizen is defined as any U.S. citizen who is residing in another country and who is not an active duty military member or dependent. If an overseas citizen is at least 18 years old, he or she is defined as an eligible overseas citizen, meaning that he or she is old enough to vote in U.S. elections. Estimates of the size and geographic distribution of these two groups (overseas citizens and eligible overseas citizens) are based on statistical analysis combining data from many different sources.
A registered overseas voter is defined as an overseas citizen who has specifically registered to vote and has requested an absentee ballot to be sent to an overseas address. All registered overseas voters are overseas citizens, but not all overseas citizens are registered overseas voters. The OCPS is a survey that was sent to a sample of overseas voters following the 2014 General Election based on voter files gathered from State and local election officials. Using the sampling weights, survey results can be generalized to the population of overseas voters, but not necessarily to all overseas citizens. There is no centralized record of all overseas citizens, so there is no way to construct a representative sample of the entire group of overseas citizens.
Q12) Who did overseas voters vote for? Are they more likely to be Republicans or Democrats?
A12) FVAP is a nonpartisan agency, and it does not collect any information about voters' party preferences or votes cast. Voter files are a matter of public record, which means anyone can see who is registered to vote and which elections they voted in, but all U.S. ballots are secret ballots. Out of respect for the privacy of overseas voters, FVAP does not ask them any questions about which candidates or issues they voted for, even when data collection is confidential and anonymous. FVAP's goal is to ensure that all military and overseas voters can complete the voting process if they desire to do so, regardless of their party affiliation, candidate preferences, or political views.
Q13) Why does the OCPA estimate for a given country differ so much from that provided by the country's government? (Added 10/18/2016)
A13) Governments vary with respect to how they define their "American" population as well as how they count it. Some governments provide a count of individuals born in the United States while others provide counts of American citizens. Even amongst those governments who provide counts of citizens, it is unclear whether dual citizens (those who are both citizens of the United States and the host country) are included in this count. Irrespective of who is counted as an "American", some governments count Americans using a census, while others develop their count using administrative records.
For the purpose of creating an estimate of the overseas voting rate, counts of the number of U.S. citizens are more relevant, as only citizens have the opportunity to vote. In addition, because there were concerns that administrative records may suffer from under coverage and/or duplicate records, counts produced using a Census are thought to be more reliable. Consequently, the OCPA estimates represent an estimate of what the government count would have been if it had been a count of the number of U.S. citizens (including dual citizens) in the country using a census. As a result of these differences in what is being measured, as well as idiosyncratic features of the country or its estimate not captured by our model, the OCPA estimate for a given country can differ substantially from the estimate provided by the country's government.
[i] Smith, C. M. (2010). These Are Our Numbers: Civilian Americans Overseas and Voter Turnout. OVF Research Newsletter, 2(4), 5-9. Retrieved from https://www.overseasvotefoundation.org/files/counting%20american%20civilians%20abroad.pdf
[ii] Government Accountability Office. (2004). 2010 Census: Counting Americans overseas as part of the census would not be feasible. Washington, DC: U.S. Government Accountability Office. Retrieved from http://www.gao.gov/new.items/d041077t.pdf
[iv] Federal Voting Assistance Program. (2005). The federal voting assistance program: Seventeenth report. Washington, DC: Department of Defense. Retrieved from https://www.fvap.gov/uploads/FVAP/Reports/17threport.pdf
[v] González-Barrera, A. (2015). More Mexicans Leaving Than Coming to the US. Pew Research Center, 19. Retrieved from http://www.pewhispanic.org/2015/11/19/more-mexicans-leaving-than-coming-to-the-u-s/
[vi] Greater detail on the relative importance of these factors can be found on page 27 of Volume 1 of the full report.