Here's another item from last week worth a second look: Is Al
Gore really leading in Michigan?
When we linked to
the latest Michigan survey from EPIC/MRA last week, we followed the lead of the
News story and gave emphasis to
the trial heat questions that included non-candidates Al Gore among the
Democrats and Newt Gingrich and (non "announced" candidate) Fred Thompson among
Republicans. The results had some Pollster readers scratching their heads, as the
Democratic question showed that Gore "would top the Democratic slate" (with 36% to
Hillary Clinton's 32%) and showed Gingrich receiving the support of 15% of Republicans.
"These results are pretty weird," wrote Pollster commenter Chris
S., "when compared against other MI polls, and other state polls around the
country." Chris has a point. When pollsters include Gore on the list of choices
of national and statewide polls, he typically receives support in the high single
digits to mid teens. Thirty-six percent is certainly high by comparison.
It turns out that EPIC/MRA asked their trial heat questions
a bit differently, and that difference most likely explains the odd result. Virtually
all of the other pollsters that include non-candidates like Gore, Gingrich and
(technically) Fred Thompson start with a trial-heat question that includes all
possible candidates and then ask respondents for their "second choice." They
can then re-allocate the second choices of Gore supporters to calculate a "vote
EPIC/MRA, on the other hand, did things differently. For
each party, they first asked voters to choose from the "announced candidates"
for president and then presented a choice from what they described as an "expanded
list" of candidates.
That their approach showed greater support for Gore is not
surprising. Presumably, many of Gore's potential backers take him at his word when he says he has no plans to run and tend to choose other candidates. However, Gore does much better when a pollster
plays "what if" and says, in essence, "imagine that Al Gore decides to run."
That is one reason why -- despite considerable internal
debate -- we have chosen to use the first question that pollsters ask on our
charts, rather than relatively hypothetical follow-up questions based on second
choice or compressed lists of candidates. For what it's worth, we are doing
some work behind the scenes here at Pollster to allow readers more choices
(such as charts based only on votes without non-candidates like Gore and
One of our fears is that as soon as we launch such a capability,
pollsters will finally decide to stop asking about non-candidates. I generally
agree with those who ask here, again and again, "why are pollsters asking about Gore?" Regardless of the way pollsters ask the vote question, the "support"
measured for Gore will be hypothetical and artificial. So why bother at all
unless he decides to run?
This result does help demonstrate something useful, however,
which has less to do with Gore than with the current support for the frontrunner
Hillary Clinton. Just how real (or solid) is her support, at least among
Michigan Democrats? The fact that preference for Clinton drops from 45% to 32% with Gore's
name included, suggests that many of her supporters remain open to an
alternative. Less clear, of course, is whether any of the actual
candidates (Obama, Edwards or any of the others) can replicate Gore's apparent appeal among Michigan Democrats (as support for the others also
drops with Gore included). But the fact that the simple addition of Gore's name
to the list of choices shakes up current preferences to this degree tells us
that the vote preferences we are watching are still tentative, and the
Democratic race is still a long way from over.
Follow Mark Blumenthal on Twitter: www.twitter.com/MysteryPollster