David Enrich
Citizens for True Democracy
January 20, 2000

The Federal Election Commission (FEC) publishes and distributes statistics and studies regarding voter registration and turnout in the United States. Researchers, students, and media trust the FEC’s publications as authoritative and reliable. A review of the independent federal agency’s methodology and data regarding trends in voter registration shows that its work is often misleading.

This report highlights two major problems in the FEC’s analyses and presentations of trends in voter registration from 1960 to 1996.

1. The FEC uses an inaccurate, incomplete, and inconsistent method to determine trends in voter registration. Its methodological error results in a substantial understatement of national voter registration as a percentage of the national voting age population, particularly in the presidential election years of 1960, 1964, 1968, and 1972. This understatement is the core of the FEC’s argument that voter registration has steadily increased for the past 40 years. Correcting the error reveals that the percentage of registered voters has not substantially changed since 1960.



2. The FEC’s website includes numerous errors, inconsistencies, repetitive data, and imprecise analyses. The inaccuracies pervade most of the Commission’s statistics on historic trends in the nation’s voting age population, voter registration, and voter turnout. The errors not only substantively affect much of the information available on the FEC’s site, but they also are indicative of the FEC’s irresponsible and lackadaisical approach to its presentation of information and its inevitable effect on students, academics, and concerned citizens.

On Monday, January 3, 2000, a brief summary of the numerous methodological and substantive errors was e-mailed to the FEC’s Office of Election Administration, and other FEC employees. The FEC has thus far taken no actions to either contact concerned parties or change or remove its misleading data from its website.

Supplemental research materials are available online via Citizens for True Democracy’s website, at www.truedemocracy.org.

The FEC considers three variables in its analysis of voter registration and turnout trends.

• Voting Age Population (VAP) is the total number of people in the U.S. who are at least 18 years old. The FEC calculates the VAP by state, as reported by the Census Bureau. The number is higher than the number of people who are actually eligible to vote, because voting age population includes non-citizens and others who, for various reasons, are ineligible to vote. In addition to natural population growth, three constitutional amendments between 1960 and 1971, and the 1965 Voting Rights Act, have substantially increased the voting age population.

• Registration is the number of registered voters. The FEC calculates the national registration total as reported by individual states and the District of Columbia.

• Turnout is the number of people who actually vote in a given election. The number is sometimes under-stated because it reflects only the number of people who vote for the highest office on the ballot. In other words, if an individual votes for a congressional, but not presidential, candidate, his vote is not included in the turnout calculation.

The inter-relationship between these variables can be deceptive. As prices are adjusted for inflation in historical analyses, so, too, must registration be adjusted for population growth to determine trends in political participation. Fluctuations in the number of registered voters are only relevant to the extent that they reflect something more than changes in the VAP. Registration is a meaningful gauge of trends in voter participation only when one considers it relative to simultaneous fluctuations in the VAP.

The FEC therefore uses two ratios to detect national trends in turnout and registration: voter registration as a share of the VAP, and voter turnout as a share of the VAP.

In analyzing trends in voter registration, the FEC examines the national voting age population (VAP) and national registration. Starting in 1960, the FEC includes all states’ (and the District of Columbia’s) voting age populations in its national totals.

For a variety of reasons, however, the FEC does not include voter registration information for every state in its calculation of the national total. In 1960, 16 states’ voter registration figures are not included in the national total, and two states lack data for every subsequent election.

The FEC lacks the following states’ registration figures:

• In 1960: Alabama, Alaska, District of Columbia, Iowa, Kansas, Kentucky, Mississippi, Missouri, Nebraska, New Mexico, North Carolina, North Dakota, Oklahoma, South Dakota, Virginia, Wisconsin, and Wyoming;

• In 1964: Alabama, Alaska, Iowa, Kansas, Mississippi, Missouri, Nebraska, North Carolina, North Dakota, Wisconsin, and Wyoming;

• In 1968: Alaska, Iowa, Kansas, Missouri, North Dakota, and Wisconsin;

• In 1972: Iowa, Missouri, North Dakota, and Wisconsin;

• From 1976 to 1996: North Dakota and Wisconsin.

Because the FEC includes the voting age populations of all states and the District of Columbia in its national VAP calculation, each state for which the FEC does not include registration information is in effect credited with zero percent voter registration. This results in a substantial under-statement of national voter registration as a percentage of the VAP particularly for presidential elections in 1960 through 1972 in particular, and it noticeably skews results for all other elections.

The FEC lacks voter registration data for North Dakota and Wisconsin for every year, yet continues to include both states’ voting age populations in the national totals. A brief examination of those states’ voter registration procedures is in order:

• North Dakota automatically registers every eligible citizen on his or her eighteenth birthday. The FEC treats North Dakota as having zero percent registration, but, by definition, the state has 100 percent registration, because every eligible citizen is registered to vote.

• Wisconsin allows "same-day" voter registration as well as conventional registration prior to the election. Thus, the state does not keep detailed voter registration information distinct from figures on voter turnout. As a share of its VAP, Wisconsin consistently has voter turnout well above national averages, and it is therefore likely that its voter registration, too, is above par; it is certainly much higher than the zero percent that the FEC implies.

One could devise an equation to estimate each missing state’s voter registration, but it would inevitably be controversial and of dubious validity. The only feasible way to achieve an accurate estimate of national registration relative to the VAP is to exclude a state’s VAP from the national total unless the FEC also has that state’s voter registration information for a given year.

This analysis reveals that the FEC’s methodology substantially skews results:

• In 1960, the FEC estimates that voter registration as a share of the VAP is 58 percent; adjusted to reflect the 16 missing states, registration jumps to 73 percent.

• The FEC says that registration in 1964 was less than 65 percent; in reality, it was approximately 75 percent.

• Voter registration as a share of the VAP wa more than 73 percent in 1968, but the FEC concludes that it was less than 68 percent.

• In 1972, the FEC says that registration was approximately 69 percent, when the adjusted figure is close to 74 percent.

• From 1976 to 1996, there is a consistent difference of 1.5 to two percentage points between the FEC’s estimates and voter registration as a percentage of the applicable VAP.

More striking, though, is what these discrepancies show about trends in registration over the past four decades.

• According to the FEC’s data and estimates:

1. Between the ten presidential elections since 1960, registration as a share of the VAP has substantially increased six times, while remaining virtually constant twice, and substantially declining only once, between 1984 and 1988.



2. Since 1960, voter registration has skyrocketed more than 16 percentage points, from 58 percent in 1960 to over 74 percent in 1996.

• Adjusting the voter registration figures to account for the multiple missing states, we find that:

1. Registration as a share of the VAP has increased four times, remained virtually constant twice, and declined three times between presidential election since 1960.



2. Voter registration’s net increase since 1960 has been less than three percent, growing from 73 percent in 1960 to 76 percent in 1996, with constant fluctuations back and forth.

By this measurement, the FEC’s estimate of the growth in voter registration is more than five times too high.

Congress established the Federal Election Commission in 1975 as the enforcement arm of the 1971 Federal Election Campaign Act. In Fiscal Year 2000, Congress allocated more than 38 million dollars to the FEC and its 350 employees. Though its primary role is the enforcement of federal campaign finance laws and the administration of public funding for presidential elections, the FEC also reports to Congress and issues press releases and reports regarding national trends in voter turnout and registration.

The National Voter Registration Act of 1993, commonly referred to as the "Motor Voter" law, instructs the FEC to periodically report to Congress on trends in voter registration. The FEC last wrote to Congress in June 1997. It summarized its findings:

This statement, which the media widely quoted and the FEC has prominently placed on its website, is problematic:

• The FEC later amended its registration estimate to 74.4 percent, reflecting additional states that provided registration information. Reporting only on states for which it has comprehensive information would improve the accuracy of the FEC’s analysis, even absent one or more states’ data.

• 72.77 percent is not the "highest percentage of voter registration since … 1960." Registration was higher in 1960, 1964, 1968, 1972, and 1984. If the FEC had adjusted its 1996 data, however, it would increase to 76 percent, which is in fact the highest registration since 1960.

Most important, the FEC’s statement to Congress, and its general presentation of data, is very misleading. Since 1960, voter registration has remained virtually constant. The increase between 1992 and 1996 barely compensates for registration’s post-Watergate decline.

The FEC’s calculation of the VAP has not changed since 1960. Between 1960 and 1972, though, major changes occurred which substantially increased the number of Americans that were eligible to vote:

• The 23^rd Amendment to the U.S. Constitution, ratified in 1961, made citizens in the District of Columbia eligible to vote in presidential elections. (The FEC includes D.C.’s voting age population in the national total in 1960, even though D.C. residents were ineligible to vote until the 1964 elections.)

• The 24^th Amendment, ratified in 1964, abolished payment of poll or other taxes as a prerequisite for participating in elections.

• The 1965 Voting Rights Act increased protection of voting and registration rights against discrimination and intimidation.

• The 26^th Amendment, ratified in 1971, lowered the voting age from 21 to 18. (The FEC includes all U.S. residents, 18 or older, in the VAP starting in 1960, even though 18 year-olds did not have a constitutional right to vote in the 1960, 1964, and 1968 presidential elections.)

The FEC says that voter registration increased by more than 18 percent between the 1960 and 1976 elections, the first presidential election year by which all four changes had fully taken place.

A closer analysis, however, reveals a different trend. In 1960, voter registration as a share of the VAP was 73 percent. In 1976, voter registration had declined to less than 71 percent. Between 1960 and 1980, registration declined by nearly three percentage points.

One can attribute this decline to a combination of ineffective policies, inadequate enforcement, and the deterioration of Americans’ trust and interest in politics following the Watergate scandal. The improved voter turnout in recent years may be a product of the increased ease with which people can register to vote, which would be a welcome development. Even if this is the case, though, the FEC’s analysis of trends in voter registration – and its prognosis for the success of laws like "Motor Voter" – is misleadingly optimistic.

Never before have so many Americans had such direct and immediate access to government records and primary source documents. The Internet’s rise as a reliable medium between citizens, organizations, and the government offers an unprecedented opportunity for an informed citizenry, and an accountable government.

The FEC is the most trusted – and often the only – source of campaign finance and voter participation data. Through various media, the FEC’s studies and statistics are quickly and widely disseminated. In addition to conventional wire accounts of FEC research, the media often refers interested individuals to the FEC’s website for additional, reliable information. For example:

• In an article titled "The Cutting Edge," the Los Angeles Times encourages readers to visit the FEC website for "registration and voter turnout figures [and] voting statistics . . ." (May 11, 1998)

• A July 27, 1998 article in the Atlanta Journal-Constitution responds to a reader’s question about where to find accurate registration and turnout statistics by suggesting specific pages on the FEC’s website.

The FEC, and its FEC.gov website, have significant name recognition, and people automatically trust its information as authoritative and reliable. The FEC therefore has a responsibility to provide the most accurate and comprehensive information possible. Its misleading methodology regarding voter registration and the national voting age population is troubling. Equally concerning, however, is the abundance of inaccurate, inconsistent, and generally distorted data that is easily accessible on the Commission’s website.

Despite its recent renovation and "streamlin[ing]" of December 23, 1999, the FEC’s website includes a considerable amount of inconsistent information. Few researchers assume that they need to crosscheck information that the FEC posts on its website. While carelessness presumably caused most of the following errors, the result is misleading and often inaccurate data that can substantially skew research’s results.

One source of the FEC’s inconsistency is that multiple pages supposedly provide the same information:

• http://www.fec.gov/votregis/turn/natto.htm ("National Voter Registration and Turnout in Presidential Elections – 1960 to 1992") includes the VAP, "total" registration, turnout, and ratios of registration and turnout to the VAP. This page does not say that the national registration information lacks numerous states’ data.

• http://www.fec.gov/pages/tonote.htm ("Voter Registration and Turnout in Presidential Elections by Year: 1960-1992") offers the VAP, registration, turnout, and registration and turnout as percentages of the VAP for individual states, and as national averages. It does not say that the national registration data lacks several states’ information.

• http://www.fec.gov/pages/htmlto5.htm ("National Voter Turnout in Federal Elections: 1960-1996") includes the VAP, registration, turnout, and turnout as a percentage of the VAP. This is the only page that notes that the national voter registration information provided is missing numerous states, but this page does not include registration as a percentage of the VAP.

The following is a small sampling of the inconsistent and inaccurate data on the FEC’s website. On different pages, the FEC lists:

• Maine’s 1960 VAP as 588,000 and 688,000;

• 1972’s national voter registration as 97,328,541 and 97,283,541;

• Kansas’s 1972 VAP as 2,223,000 (the exact same as that of Kentucky, which directly follows Kansas alphabetically) and 1,539,000;

• Maryland’s 1972 VAP as 2,271,000 and 2,721,000.

The above data and pages are within two mouse "clicks" of FEC.gov’s front page.

The following ratios, which the FEC provides via its website, reflect the FEC’s irresponsible disregard for accuracy:

• In 1960, West Virginia’s registration is 101.4 percent of its VAP;

• In 1964, Minnesota’s registration is 108.3 percent of its VAP; West Virginia’s is 100.6 percent;

• In 1968, Minnesota’s registration is 103.3 percent of its VAP;

• In 1972, Minnesota’s registration is 100.3 percent of its VAP;

• In 1996, Maine’s registration is 106.0 percent of its VAP.

Clearly, these are statistical impossibilities. They could reflect what the FEC refers to as "deadwood," registered voters who have moved, died, or registered under a different name or address. They could also reflect data that was entered inaccurately into FEC spreadsheets. Regardless, the FEC lists the above percentages without explanation or recognition of their invalidity.

This analysis of FEC methodology, statistics, and data presentation is instructive on two, distinct levels. First, the recent growth in voter registration is not part of an historic trend. Since 1960, voter registration has barely kept pace with the nation’s increasing population. Increased registration since 1992 could well represent nothing more than a minor blip in long-term trends. As in 1960 and 1980, registration can sharply increase only to decline again ten years later.

Moreover, the 1960s and early 1970s witnessed several major reforms in voting procedure that were specifically designed to increase voter registration. Their collective failure is cause for concern. It also should be cause for skepticism about the "Motor Voter" law’s realistic potential for substantially improving voter registration, when four dramatic policy changes could not.

Second, voting statistics that appear on the FEC’s website are often inaccurate and inconsistent with other statistics elsewhere on the site. Many tout the information revolution’s empowering characteristics, but if misinformation reigns, the revolution fails. The FEC has a reputation as a reliable source for campaign finance and voter participation statistics and information; countless students, researchers, and concerned citizens visit its site every day, and do not think twice about the accuracy and reliability of the information that they find. Too frequently, that information is false and misleading.

TOP OF PAGE

REPORT INDEX