By now many of you know about a new outbreak of nasty phone
calls in Iowa.
I have to say that this episode truly puzzles me. Politico's Ben Smith reported
the key details yesterday:
Two bloggers today reported receiving calls last
night from a pollster testing whether John Edwards' failure to drop out to take
care of his ailing wife could damage his campaign.
The pollster asked whether
"desmoinesdem," a well-regarded liberal Iowa blogger, would not support Edwards
because "he chose to continue the presidential campaign instead of staying
home with his wife who has cancer," the blogger reported. A blogger
on John Edwards campaign website, doridc, shared a similar
The two blogger/respondents also said the "survey" included
negative statements about both John Edwards and Hillary Clinton, which led both
to believe that the calls came from someone associated with the Obama campaign.
My initial reaction was that these calls were part of some
sort of message testing rather than a so-called "push poll," and that a real
message testing poll could have involved any
of the Democratic campaigns. As regular readers know, a "push poll" is not a
poll at all but political telemarketing -- usually an effort to spread a nasty
negative message -- under the false guise of a legitimate poll. However,
campaigns sometimes use real surveys to test potential negative messages for
advertising. I've written more about the differences here; see also Stu
Rothenberg's piece and the AAPOR statement on "push
If this were a real "message test," the absence of
statements about Obama or any of the other Democratic candidates does not
necessarily implicate either Obama as the sponsor (or Dodd, Richardson or
Ambinder made a similar point yesterday). As a campaign pollster, I always
included "negatives" on my own client whenever I tested potential attacks on
the opposition. The idea was to try to keep some semblance of fairness and
balance in the questionnaire (so respondents would not immediately hang up in
anger) and to try to simulate the effects of the likely exchange of attacks
that would occur if the campaign "went negative." For what it's worth, I know
that Harrison Hickman, the Edwards pollster for whom I once worked, has always
taken the same approach.
Ben Smith, who has been all over this story for the last 24
a real call center that uses the name "Central Research" in New York and
another Central Research in Arkansas, and speculated about projects they have
done in the past. However, he subsequently reached the people that run those
companies and each flatly
denied any involvement. He also reports that each of the Democratic
campaigns - Obama, Clinton, Edwards, Biden, Dodd and Richardson - "all quickly
denied this was their poll."
Puzzled, I went back and re-read the reports of the first two
blogger/respondents (as well as another respondent account Smith reported
this morning). All tell essentially the same story:
- Both report being asked first about their voter preference in the Democratic caucuses, about the strength of their support for their first choice and about their second choice.
- Both report being asked just two more questions: A three-way forced choice question asking them to choose from among three statements as the "most important reason not to support Clinton." They were then asked a similar question with two negative statements about Edwards.
then, both bloggers report (or at least imply) that the interview ends abruptly..
"End of survey" says desmoinesdem. "There
were no questions about what issues are you interested in," writes dorisdem, "or
even how likely are you to caucus." Neither mentions being asked demographic
questions (what is your age, race, etc.), although is it possible both simply left
that part out.
If the "poll" included just the five questions above, it fits the profile of
the a real so-called push poll,
again, not a poll at all but a negative "advocacy call" masquerading as a
"negative statements" are also strange, and not just because of the outrageous
and incendiary reference to Elizabeth Edwards' fight against cancer. The two
bloggers report essentially the same statements. This is dorisdem's memory:
[They ask] why do you think Hillary Clinton is a
weak candidate and gives 3 choices. A) Is a weak general election candidate.
B)Is too dependent on lobbyist money. C) Won't bring change.
Then why do [you] think John Edwards is a weak
candidate with 2 choices A) a weak general election candidate because his
positions are too liberal B) He should be home with his wife who has cancer.
are just not the sort of statements that I can imagine any of the campaigns
wanting to "test" in this form at this stage of the campaign as potential
fodder for television or direct mail advertising. Think about the ways the
campaigns are criticizing each other now in speeches, online videos and
debates. The statements in the calls make no reference to votes on Iraq, Iran,
trade policy or double-talk
in regards to Senator Clinton; nothing about hypocrisy, being too negative or
"piling on" in regards to Senator Edwards.
oddly constructed questions look mostly to me like a clumsy attempt to dress up
as a "poll" the beyond-the-pale reference to Elizabeth Edwards' illness.
very least, I find it utterly inconceivable that Harrison Hickman or the
Edwards campaign had any connection to a five-question survey of this sort, and
extremely implausible that it was part of any real poll conducted by anyone
who would be doing this?
bottom line is that I have no idea who is behind it, but we ought to consider
another scenario as at least as plausible as the notion that this came from one
of the Democratic campaigns. It is also possible that this was the work of some
independent group with Republican ties that sees some value in gathering crude
information about the Democratic race while fomenting ill-will and infighting
in the Democratic ranks. If that was the goal, you need only read the comments
under the posts of the blogger/respondents I linked to above to see evidence of
just that happening.
again, we really have no idea. It could be anyone, and we'll probably never
Just to help clarify the record, blogger/respondent doridc makes
two statements that are not quite right. First:
In polls from reputable sources they never ask for specific
voters as was the case here.
true. Most surveys that sample from registered voter lists, including just
about all of the campaigns, ask for the name on the list when they call. They
can randomly select an individual in the household (from the list). Asking for
a specific person by name allows them to match up the answers to actual vote
history on the list for analytical purposes.
Also they used my voter id as listed in the Iowa Democratic
Party database because my full name does not fit, it is missing the last 3
letters. So it was a candidate driven call.
necessarily. As I understand it -- and I have this information from individuals
with firsthand knowledge of the process -- the list that the Iowa Democratic
Party sells to campaigns is built by appending caucus "vote history" data that
they collect to the registered voter list provided by the Iowa Secretary of
State. I may be wrong on thus, but I assume that if doridc's name is truncated
in the file, that truncation occurred on the Secretary of State's list that is
available to anyone.
Smith links to this post along with confirmation oF one point above:
After talking to Blumenthal today, I went back to one of the
respondents, "desmoinesdem," to ask her about some of these details: Was
she asked about whether she plans to vote, or about her age or party
affiliation. She wasn't.
Desmoinesdem has also updated her original post with another interesting observation:
In the comments, yitbos96bb suggested a possibility that hadn't occurred to me. The pollster may be testing negative messages against Hillary (the front-runner) and whomever the respondent supports. So doridc and I got the negative messages about Edwards, but perhaps if we had named a different candidate as our first choice, we would have gotten the questions about Hillary and our first choice. A Republican group paying for a poll like this might be testing to see what kind of messages would work best against Hillary and whomever Democratic respondents lean towards.
Update (11/20): More details here.