The 2006 release of the National Election Study, a large biennial national survey, has just been made available to the public.
As you might expect, I quickly went into the question listing to see how they queried respondents about early voting. Along with Mike McDonald and Thad Hall, I am attending an election reform conference next week in Columbus, OH, and am working on a paper on early voting and turnout.
So I was disappointed to find out that, in the 2006 study, the NES did such a poor job asking about early voting. Specifically, they ask these items:
QUESTION:
——–
[Did you vote in person on election day, or did you vote
by mail? / Did you vote in person on election day, or did
you mail in an absentee ballot before election day?]
VALID CODES:
———–
1. Voted in person
2. Voted by mail
Most readers of this blog will recognize the problem with this question: what about voters who can vote early in person (comprising, for example, 20% of the Florida electorate)? What if people carried in their absentee ballot (admittedly a very small proportion of the electorate in any state)? What about Oregon, and increasingly Washington, where there you may “vote by mail” on election day?
What’s surprising is that the NES did such a better job in 2000, 2002, and 2004. Here are the items they used that year:
P001245 C4. Did R vote on election day or before
C4. IF R VOTED:
Did you vote on election day — that is, November 7th
2000, or did you vote at some time before this?
———————————————————————
1. ELECTION DAY –> SKIP TO C5
5. SOME TIME BEFORE THIS –> C4a
8. DK –> SKIP TO C5
9. RF –> SKIP TO C5
0. NA; INAP, 1-3, 8, 9, 0 in C1
==============================
P001246 C4a. How long before election did R vote
MD1: EQ 0, MD2: GE 98
Numeric
C4a. IF R VOTED PRIOR TO ELECTION DAY:
How long before November 7th did you vote?
———————————————————————
[PROBE: A FEW DAYS, A WEEK, LONGER THAN THAT?]
01. Less than one week, 1-6 days
02. One week; 7 days
03. 1-2 weeks; 8-14 days
04. 2-3 weeks; 15-21 days
05. 3-4 weeks; 22-28 days
06. One month; 29-31 days
07. More than one month; 32-60 days
11. A few days; a couple of days; several days — NFS
12. A few weeks; a couple of weeks; several weeks — NFS
91. More than a few days — NFS
92. More than a few weeks –NFS
97. Other
98. DK; not sure
99. RF
00. NA; INAP, 1, 8, 9, 0 in C4
==============================
P001247 C4b. Did R vote in person or absentee
C4b. IF R VOTED PRIOR TO ELECTION DAY:
Did you vote in person or by absentee ballot?
———————————————————————
1. IN PERSON –> SKIP TO C5
5. ABSENTEE BALLOT –> SKIP TO C5
7. R VOLUNTEERS: BY MAIL [OREGON ONLY] –> SKIP TO C5
8. DK –> SKIP TO C5
9. RF
0. NA; INAP, 1, 8, 9, 0 in C4
Now admittedly, they still don’t have the “in-person early voting” quite right, but at least they give the respondent more (and more accurate) options.
Perhaps I am seeing only the early release of the 2006 survey, and the full version will have the complete series. But at this point, I sure hope that the NES doesn’t make the same mistake in 2008, and instead returns to the 2004 version of the item (and adds the “early voting in person” option).
An additional problem: the NES asks respondents whether they voted “in this county” or in a different county, presumably to track respondents who moved since the time of the election and the date of the interview. But how are respondents in states with regional voting centers supposed to answer that item?