Mark Blumenthal | January 11, 2007
Topics: Instant Reaction Polls , Iraq
Two survey organizations - CBS News and ABC/Washington Post - went into the field immediately after the Presidential address last night to gauge reactions among those at home. Both polls appear to show the same polarization in reaction: Republicans are supportive of the President's "troop surge" plan, most independents and virtually all Democrats are opposed. As always, we need to be cautious about instant reaction polls. Opinions may change in response to the news coverage over the next few days, but despite some differences that may seem divergent when taken at face value, both surveys paint roughly the same general picture.
The two polls differed in terms of their methodology. CBS News conducted a "panel-back" survey. They attempted to interview respondents from a larger survey conducted last week for a second time. They were able to contact and interview 458 adults and (I assume) weighted the results as they usually do to match national US Census estimates for demographic characteristics like gender, age and race. The ABC/Post survey involved a fresh new random sample of 502 adults, all contacted for the first time immediately after the speech.
The biggest challenge of fielding this sort of "instant reaction" poll following a presidential address is that the president's fans are more likely to tune in than other Americans. Did that happen in this case? The evidence is mixed. The two polls produced estimates of the audience size that differed, but not by much (31% on the CBS poll and 42% on the ABC/Post poll).** However, the Post's summary notes that "the President's supporters were disproportionately represented among the audience," while CBS found few differences between speech-watchers and other Americans. According to the CBS release, "Democrats, Republicans and Independents were about equally likely to have watched."
That difference probably explains the divergent results among speech watchers. According to the Post, "47 percent [of speech watchers] support sending more troops, while 51 percent oppose." On the CBS poll, 33% of speech watchers favor more troops and 59% oppose."
When you look at all adults, however, the polls show more similar results. Both polls show similarly strong polarization, with most Republicans favoring a troop surge, and most independents and Democrats in opposition (thanks to Jon Cohen at the Washington Post for providing full cross-tab results from their survey in the table below).
I am reading between the lines a bit, but the data above suggest that general assessments of President Bush- both among speech watchers and other Americans - are driving judgments about the troop surge. Since the majority of Americans are skeptical of Bush, they are also skeptical of this new proposal. I would guess that if we tabulated these results by the Bush job rating, we would see an even greater polarization: Those who approve of Bush's job performance overwhelmingly in favor, while those who disapprove are overwhelmingly opposed.
**My calculation suggests that the difference in the estimate of the audience size is statistically significant, but keep in mind that the other differences in methodology (a "panel back" survey vs. a fresh random sample) may well explain that difference.
Also keep in mind that on a one-night survey of this sort, the pollsters
must abandon the usual "call back" procedures designed to interview those who
are harder to contact. In this case, the
challenge is especially acute in the Eastern time zones, where the pollsters
did all of their dialing between 9:30 and 10:00 p.m. So it is likely that both samples - even
after demographic weighting - are skewed a bit toward those more likely to stay
at home. Does this methodological compromise skew the
substantive results? Pollsters will debate
that point, but one reassuring bit of evidence is that the CBS post-speech
sample of adults had roughly the same party identification result (35%
Democrat, 29% Republican) as the larger pre-speech
debate sample (35% Democrat, 27%