Earlier today, the Pew Research
Center released results
from their latest national survey, which provides must
read analysis on the state of perceptions of the 2008 race for president. However,
an "outtake" cross-tabulation from that report sheds some new light on whether
the loose
screens used by national polls to report national presidential primary
preferences may be distorting the results. Consider, for example, the turnout
in next year's Democratic primaries and caucuses is unlikely to amount to more
than 10% of the adult population, yet most national surveys report on the
preferences of the 35% to 55% of adults (or of registered voters) that identify
or lean to the Democrats. The key question we have been asking is whether that
discrepancy makes a difference in the results?
The Pew survey has two helpful characteristics in this
regard. First, they have asked the same vote and demographic questions on their
last two surveys, allowing larger than usual sub-samples of Democratic
(n=1,188) and Republican (n=1,059) identifiers or "leaners" that are registered
to vote. Second, both surveys include a question generally considered predictive
of voter turnout:
How much thought, if any, have you
given to candidates who may be running for president in 2008, a lot, some, not
much or none at all?
I emailed the analysts at the Pew Research
Center and they shared the
following table, something apparently prepared for their report but cut from
the final draft:
Two findings stand out: Among Democrats, Barack Obama gets a
higher percent of the vote (27%) among those paying a lot of attention or
paying some attention (28%) than among those paying little or no attention
(19%). Similarly, Rudy Giuliani gets a higher percentage of the vote (36%)
among Republicans paying a lot of attention or some attention (34%) than among those
paying little or no attention (28%). Conversely but not surprisingly, in both
cases, those paying the least attention are also the most likely to tell
pollsters they are undecided.
So what do these results say about Chris Bowers' theory
that national polls are overstating Hillary Clinton's lead? The evidence here
is mixed, at best. Obama certainly does better among more attentive voters,
although that finding is not particularly surprising given his rapidly growing
name recognition in recent months. However, Clinton also
does better among the most attentive Democrats. Thus, her margin over Obama
among those who pay "a lot" of attention (11 points in the combined March/April
data) is actually a few statistically insignificant points higher than her
margin among all Democrats (9 points in March, 10 points in April).
Now, some cautions about the above. First, those who say
they pay a lot of attention to the candidates are more likely to vote than
those who do not, but this measure is far from a perfect turnout predictor. Pollsters
that use attentiveness to select likely voters usually do so in combination
with other measures, such as reports of past voting or future likelihood to
vote. Second, an interesting twist: Among Democrats, Al Gore does better among
the least attentive (17%) than among the most attentive (10%). Reallocating Gore's
vote using Pew's second choice question might change these numbers slightly,
but probably not dramatically.
Meanwhile, over at MyDD, Chris Bowers looks at the
Pew results showing Clinton running better among Democratic identifiers
than among independents who lean that way, and he sees an "an important structural
flaw" in his theory:
I still believe that Obama probably
does much better relative to Clinton
among voters who are paying very close, or somewhat close, attention to the
campaign than among voters who are not paying much attention at all. However,
closed primaries in several large February 5th states might cancel out that
advantage, since Obama performs relatively better among Democratic-leaning
independents who won't be able to vote in closed primaries than he does
among self-identified Democrats (in this case, I am assuming self-identified
Democrats are more likely to be registered Democrats than are
Democratic-leaning independents).
Note that several other surveys have confirmed
that Clinton's margin is narrower among independent leaning Democrats.