Harris Interactive released a new survey yesterday, based on over 2,000 interviews conducted online that focuses on the candidacy of Hillary Clinton and suggests she faces a big challenge should she win the Democratic nomination:
[H]alf of U.S. adults say they would not vote for Senator Clinton if she was the Democratic candidate, while only 36 percent say they would, with 11 percent unsure.
Some will question the Harris survey results, as they are based on a non-probability sample drawn from its online panel of volunteer survey respondents. I must admit that my first thought was to wonder whether their online panel might include a disproportionate number of "net roots" Democrats who - as we know from some research I helped conduct last year - tend to be more hostile to the Clinton candidacy. However, on closer examination, the Harris result appears to be at least not wildly inconsistent with similar measures on other recent national polls.
Here are the text and complete results of the question that produced the lead of the Harris release:
If Hillary Clinton was the Democratic nominee for President, which is closest to the way you think?
15% - I definitely would vote for, her
22% - I probably would vote for her
11% - I probably would not vote for her
39% - I definitely would not vote for her
11% - I wouldn't vote at all
Now consider the results of similar questions asked on two
recent national polls:
Pew Research Center
(1,509 adults, February 7-11, 2007) - Have you heard of Hillary Clinton, or
not? (IF HAVE HEARD, ASK:) How
much of a chance is there that you would vote for Hillary Clinton if she is a
candidate for president in 2008 - is there a good chance, some chance, or no
98% (1,408 adults) had heard of Hillary Clinton
Among those respondents:
32% good chance
24% some chance
40% no chance
4% (volunteered: don't know/refused)
Fox News/Opinion Dynamics (900 registered voters, February 13-14, 2007) - Now I am going to read a list of possible candidates for the next presidential race. For each one, please tell me whether you would definitely vote for that candidate for president, if you might vote for that candidate, or if under no conditions would you vote for that candidate.
18% definitely vote for
34% might vote for
44% under no conditions vote for
3% (volunteered: don't know)
While each of these questions asks about the degree to which
respondents are considering supporting Clinton,
they are very different, both in terms of the wording and the categories. Probably
the most important difference is that the Pew and Fox questions present three
categories, while the Harris survey presented four (plus "would not vote"). The
Pew and Fox items also used a soft positive choice ("might vote for" or "some
chance") as their middle category. Neither presented the soft negative choice
("probably would not vote for") included on the Harris survey.
Also, the Fox survey sampled registered voters, while the
Pew and Harris survey aimed to project the attitudes of all U.S. adults.
But for all their incomparable elements, these questions produce some consistent findings: 39% to 44% say they have ruled out
Also, the 15% of adults who say they
would "definitely" vote for Clinton in the
Harris survey is reasonably close to the 18% of registered voters who say they
will "definitely" support Clinton
on the Fox survey.
Putting aside the various sampling issues, the idea that
some voters could say they "might vote for someone" on one question but
"probably not" on another is not surprising, particularly when only "might vote
for" or "some chance" is the only middle option. After all, the Republican
nominee is unknown, most expect to scrutinize the candidates more closely before
voting and the election is still 18 months away.
Also, before making too much of this result, we really need
to see the Harris results in a larger context. Specifically, how do the other
Democrats (and Republicans) compare on the same measure. For example, on the
Fox News poll, Clinton's
18% "definitely vote for" score was greater than those of the other candidates.
Her "under no conditions" score (44%) was about the same as that received by of
John Edwards (45%) and within range of John McCain (40%).
What would help resolve the sampling question - at least with regard to this result - would be
a side-by-side comparison that asks the Harris question on both an online and
traditional telephone poll conducted at the same time. It is possible Harris
Interactive has already done so, as they frequently conduct parallel surveys
(online and on the phone) as part of their "propensity weighting" technique. I
sent an email to Harris to check, and will report back if and when I hear more.