Statistic and source needed. What was 2004 National undervote rate?

[L. Coyote]

Anyone have this? I have an approximate only, and the hard data to do the math is elusive.

[OnTheOtherHand]

1. yeah, I'm not aware that this statistic exists, but...

one starting place is the 2004 Election Day Survey, especially chapters 7 on "drop-off" (aka "residual votes") and 8 on undervotes and overvotes.

Chapter 7 says, "In the 2004 election, reported drop-off was 0.99 percent, the lowest level in post-World War II elections." However, clearly that isn't the Right Answer, as the chapter also notes that many jurisdictions reported implausible zero drop-off or even negative drop-off.

I don't think Charles Stewart even reports an overall statistic, but if you cross-referenced his spreadsheet with the EDS data, you might get a more accurate result than either. (Someone else may have done this work, but I don't think I've ever seen it.)

[L. Coyote]

3. THANKS, you found it:

http://www.eac.gov/clearinghouse/docs/eds2004/eds-2004-part-2-chapter-3/attachment_download/file

Table 3a. Electoral Drop-Off Rates, 1948–2004 Number of States Reporting Voter
Year = 2004
Actual Voter Turnout = 105,357,390
Highest Office Turnout = 104,322,648
Turnout* Drop-Off Rate = 0.98%


"3,688 Nevada voters, or 0.44 percent, choose “None of these Candidates” in the 2004 presidential race. Although that choice in Nevada is generally considered a “candidate” in the traditional sense of the word, Nevada’s choice suggests that in states where voters do not have a similar choice, many abstain from the presidential election, but may vote for another office on the same ballot."

[OnTheOtherHand]

4. I wonder why that figure doesn't quite match the other

Goes to show how squishy these numbers can be, I guess.

[L. Coyote]

5. The details are interesting reading. We need national standards for vote reporting

that includes a requirement for usable formatting. I paid big bucks for the massive NM 2008 Access database, and it is quite useless unless you want to work for weeks to convert it to a proper document for analysis.

Some jurisdictions report the vote totals without mention of the number of voters, or list the number of voters as the number of Presidential votes counted. This skews any attempt to define undervotes. A good standard would cure this stuff.

[OnTheOtherHand]

6. reporting standards turn out to be crucial for post-election auditing

(I know I'm not surprising you) -- so the subject has come up a lot lately, in various forms. Defining and reporting undervotes (or turnout) is one piece of it, but providing timely and usable low-level results is a big project that most jurisdictions botch.

I think Minnesota may do it best. They have precinct-by-precinct results in close to real time on election night. It's a little bit of a hassle, but they use well-defined text-delimited formats. Sounds like NM is almost the opposite of MN ;). But at least, I surmise, there is a single database. In some states you would have to go county by county.

[WillE]

2. See Greg Palast

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x434675