|
|
|
|
|
|
|
CTD E-mail to FEC's Office of Election Administration(No response as of January 19th 2000)January 3, 2000To Whom It May Concern: I have been researching ways to improve voter registration and turnout, and I encountered some questionable information on the Federal Election Commission's website. On its website, the F.E.C. provides the following information for individual states: voting age population, voter registration, and election turnout, as well as some ratios of those numbers (e.g. turnout over V.A.P., registration over V.A.P., etc.), for presidential election years between 1960 and 1996. The site also lists the national totals for each category. (Data for 1960-1992 are online at: http://www.fec.gov/pages/tonote.htm. 1996 is online at: http://www.fec.gov/pages/96to.htm.) The F.E.C. never includes Wisconsin or North Dakota's registration information. (I do not know why Wisconsin's is missing; North Dakota does not register voters.) From 1960 to 1972, though, the F.E.C. is missing several other states' voter registration information: in 1960, there are 14 states without registration info, in addition to Wisconsin and North Dakota. In 1964, the F.E.C. does not include nine states' registration figures (plus Wisconsin and North Dakota); in 1968, four are missing; in 1972, only two -- Iowa and Missouri (and, of course, North Dakota and Wisconsin) -- are missing. The F.E.C. lists voting age population numbers for all states and all years, including Wisconsin, North Dakota, and the other 14 states that are missing registration data at various points. When it computes the national total of the voting age population, the F.E.C. includes those 16 states' information (and Washington, D.C.'s). This is problematic because when the F.E.C. computes the national totals for voter registration, it does not include information for the states (and, in 1960, D.C.) that it does not have. To determine national voter registration trends, the F.E.C. calculates the ratio of registered voters to the voting age population (registration/V.A.P.). The F.E.C.'s website lists the ratios for each year, using 1960 as a baseline. By the F.E.C.'s analysis, registration as a share of V.A.P. in 1960 was below 60 percent. This methodology seems misleading because, in effect, the F.E.C. treats all states for which it does not have voter registration information as having 0 percent registration, which is not the case. I entered all of this data for 1960-1996 into a spreadsheet to see what would happen if the F.E.C. analyzed the data differently. I found that if the F.E.C. did not include a state's voting age population in the national total unless the F.E.C. also had that state's voter registration numbers, 1960's registration as a percentage of V.A.P. jumps to over 73 percent (more than 14 points higher than the F.E.C.'s estimate). The problem is most acute in 1960, because the F.E.C. does not have information for several large states, but its methodology also dramatically affects the totals for 1964-1972. In addition, the simultaneous inclusion of Wisconsin's V.A.P. and exclusion of its registered voters causes a one or two percentage point understatement for 1976-1996's national ratios (Wisconsin has several million people of voting age). Every year, the F.E.C. includes North Dakota's one-half million voting age people in its national V.A.P. estimate, but counts North Dakota's number of registered voters as zero, thereby lowering the national registration average. This, too, seems misleading, because North Dakota's voter registration policy mandates 100 percent registration, despite technically not registering anyone. (Statistically, this F.E.C. method only results in a change of about one-half of one percent.) Through the data (and graphs) presented on its website, the F.E.C. says that national voter registration has constantly, albeit gradually, increased since 1960 (from less than 59 percent in 1960 to more than 74 percent in 1996). If one only includes states with both V.A.P. and registration numbers for a given year, however, the trend becomes more or less flat: 73 percent in 1960 increases to 76 percent in 1996, fluctuating up and down (between 70 and 76 percent) the entire way. (The F.E.C. issues press releases and congressional testimony based on its analysis of registered voters as a percentage of V.A.P. For example, it says in June 1997, that: "States reported a total of 142,995,856 registered voters nationwide for 1996, amounting to 72.77 % of the Voting Age Population (VAP). This is the highest percentage of voter registration since reliable records were first available in 1960." http://www.fec.gov/votregis/nvrasum.htm. This particular statement was widely quoted by media and the government.) Despite its recent renovation (of December 23), the F.E.C.'s website includes inconsistent and sloppy data, in addition to the above methodological problem. On one page, the F.E.C. says that total national voter registration in 1972 was 97,328,541; on another, it says that it was 97,283,541. On one page, the F.E.C. says that Maryland's 1972 V.A.P. was 2,721,000; on another, it says that it was 2,271,000. On one page, the F.E.C. says that Kansas's 1972 V.A.P. was 2,223,000 (the exact same as Kentucky's, which directly proceeds Kansas alphabetically); on another, it says that Kansas's V.A.P. was 1,539,000. On one page, the F.E.C. says that Maine's 1960 V.A.P. was 588,000; on another, it says that it was 688,000. (There are two main sources for the "same" data on the F.E.C.'s site. The (more or less) correct data is at: http://www.fec.gov/pages/tonote.htm. The frequently incorrect data is at: http://www.fec.gov/votregis/turn/turn.htm. Both pages are easily accessible via http://www.fec.gov.) The F.E.C. also says that in the past 40 years numerous states, most frequently Minnesota and West Virginia, had registration that was more than 100 percent of that state's V.A.P. (For example, Minnesota and West Virginia in 1964: http://www.fec.gov/pages/rat64.htm.) Also, the F.E.C. and Census Bureau's national V.A.P. totals rarely coincide, with a typical difference of several million people. (For example, see (again) 1964: http://www.fec.gov/pages/rat64.htm versus http://www.census.gov/population/socdemo/voting/history/vot07.txt.) These inconsistencies, inaccuracies, and methodological problems are troubling because any researcher assumes that information that the F.E.C. presents is reliable, and, in numerous cases, it is not. Please let me know what happens with the various problems that I have addressed. You can contact me via this e-mail address or the contact information below. I would be happy to e-mail you the spreadsheets that I have been using, if they would be helpful. Sincerely, ______________ States without voter registration data:
The Census Bureau's page of voting age populations, from 1964 to 1996:
______________
|
