Over the weekend, managers for two Republicans presidential candidates
had exchanged poll memos in an act that has become a regular feature of the
race so far. This time, as reported by the Politico's
Jonathan Martin, it was Brent
Seaborn of the Giuliani campaign and pollster John
McLaughlin on behalf of the nascent Fred Thompson campaign opining on the
meaning of recent national polls.
Needless to say, both make the most of the poll results that
favor their candidate. For "Team Rudy," that
means emphasizing national results showing Giuliani leading. For the Thompson
campaign, it means highlighting polls showing dissatisfaction among the rank
and file Republicans with the "current GOP crop" of candidates (I'm a
little surprised that Team Fred didn't make more of the recent Gallup analysis showing
Thompson leading Giuliani nationally among those Republicans familiar with the
top four candidates...but what do I know). As always, read these memos with the appropriate
skepticism - their authors will understandably make the best case that they
can.
I thought the McLaughlin/Thompson memo especially link-worthy
due to the intriguing collection of past exit poll data from Republican contests in Iowa,
New Hampshire and South Carolina (links added):
Iowa -- 1996
Decided Who To Vote For In...
Last 3 Days 23%
Last Week 19% (42% cumulative)
1996 CNN/VNS Exit Poll Data
New Hampshire -- 1996
Decided Who To Vote For In...
Election Day 23%
Last 3 Days 20% (43% cumulative)
Last Week 22% (65% cumulative)
1996 CNN/VNS Exit Poll Data
Decided who to vote for in...
Election Day 14%
Last 3 Days 12% (26% cumulative)
Last Week 24% (50% cumulative)
2000 CNN Exit Poll Data
South Carolina -- 1996
Decided Who To Vote For In...
Today 17%
Last 3 Days 14% (31% cumulative)
Last Week 24% (55% cumulative)
1996 CNN/VNS Exit Poll Data
When Did You Decide To Vote?
Today 9%
Last 3 Days 10% (19% cumulative)
Last Week 19% (38% cumulative)
2000 CNN Exit Poll Data
And for those curious, I gathered the results from the same
questions taken from exit polls conducted for the recent Democratic primaries and caucuses in
these same states (all NEP exit or entrance polls as posted by CNN):
Decided who to vote for in...
Last 3 Days 21%
Last Week 21% (42% cumulative)
Last Month 27%
Before that 30%
Decided who to vote for in...
Last 3 Days 35%
Last Week 19% (54% cumulative)
Last Month 19%
Before that 26%
Decided who to vote for in...
Election Day 19%
Last 3 Days 14%
Last Week 21% (54% cumulative)
Last Month 26%
Before that 20%
Decided who to vote for in...
Election Day 15%
Last 3 Days 11%
Last Week 21% (47% cumulative)
Earlier This Year 34%
Last Year 18%
The bottom line: We need to distinguish between voter preferences and their final decisions. The polls we are seeing
now measure current preferences, but those early leanings can change. Historically, these
exit poll data tell us that roughly half the voters in Iowa,
New Hampshire and South Carolina have waited until the last week to make their final
decisions. McLaughlin is right: "It is foolhardy for any campaign to suggest
that voter positions are already solidified at this stage of the cycle."
UPDATE - Chris "The Fix" Cillizza reports a contrary view:
As Washington Post polling director Jon Cohen pointed out to The Fix, most respondents to these exit polls may not want to admit that they made up their minds months ago, instead of at the end of the campaign after thoughtful consideration of the issues and the candidates. Second, the way the exit poll question is asked tends to favor late-deciding responses. The options often offered to respondents are "today," "in the last three days," "in the last week," "in the last month" or "before that" -- three "late-decider" options and two "early decider" options, a fact that may subtly influence voters to choose a late option even if they made their decision much earlier.
Consider also that a lot of those final late "decisions" were to stick with the candidate the voter preferred all along. Having said that, and allowing for the possibility of the sort of measurement error that Cohen talks about, these exit poll results are still generally consistent with this history of these early primaries, in which voter preferences sometimes shift significantly in campaign's final weeks (see, for example, the NCPP analysis linked to below by commenter Andrew).
Finally, note that both Cillizza and Ryan Sager (via Kaus) point out that while the Thompson/McLaughlin memo cites a Diageo/Hotline poll showing Republican discontent with the GOP field, the same survey included Thompson as a candidate. The question about dissatisfaction with the candidates followed the trial-heat ballot.
Typo corrected.