Tell a Republican operative that the Bush administration should have foreseen Sept. 11, and he’ll lecture you about the difficulty of prediction. To know exactly what will happen, he’ll argue, you have to collect all the relevant information—and all too often, you don’t know which information is relevant until the event has happened. Ballot Box agrees with this analysis. But ask the same Republican operative about Democratic criticism of the administration on this issue, and he’ll predict the political fallout with confidence: Democrats will lose because polls show the public trusts President Bush.
Democrats seem to be buying this prediction. According to today’s Washington Post, Democrats are toning down their criticism of Bush’s handling of intelligence about Sept. 11 since their strategists “say privately the party can’t win a political fight with President Bush on terrorism and need[s] to push the debate back to domestic issues.” Democratic pollster Mark Mellman told the Post, “There will be little to be found out that will cause people not to trust [Bush] to deal with” terrorism.
Ballot Box doesn’t get it. If you couldn’t have predicted Sept. 11 from the limited information Bush and his intelligence agencies had beforehand, then you probably can’t predict the politics of that issue from the limited information we currently have about what Bush and his agencies knew.
Why are the Democrats falling for this fatalism? Probably because, as the Post indicates, it’s coming from pollsters. People think polls are scientific. If a poll suggests you can’t win, you’re supposed to give up. But let’s take a look at some of these polls. On May 17, Bob Novak, a co-host of CNN’s Crossfire show, warned on the air:
Before the Democrats get too overexcited about how they’re really going to make a lot of gains on this little story, whatever it is, I’d like to read you a poll. … This is a CNN/USA Today/Gallup poll. Should there be a congressional investigation into when the administration released information on 9/11 warnings? Yes, 43 percent; no, 55 percent. … You don’t even have a majority in the country for a congressional investigation.
Hold it right there. One of Ballot Box’s rules for being an intelligent political consumer is never to accept a description of what a poll says without seeing the questionnaire. Here’s the question to which Novak was referring: “Finally, do you think there should or should not be a congressional investigation into the fact that the Bush Administration did not release this information sooner?”
Notice that the proposed investigation is confined to the timing of the disclosure of what the administration knew. Some poll respondents who think there should be an investigation of what the administration knew (not just when it disclosed this knowledge) are likely to say no to this proposal. As a result, they’re counted in the post-poll spin as people who oppose any investigation.
The fallacy doesn’t end there. Even if an inquiry begins with limited scope, it can soon reshape public opinion about what its scope should be. Investigation changes our understanding of what we should be investigating. That’s one lesson of our failure to foresee Sept. 11: If the FBI had launched a nationwide investigation of flight schools based on the now-famous Phoenix memo, somebody might have known to look for Mohamed Atta before it was too late. From a Republican perspective, the Whitewater probe illustrates the same dynamic. An investigation of why the Bush administration failed to disclose what it knew before Sept. 11 could easily transform itself and public opinion in the same way.
The simpler way in which investigations change polls is by filling in the blanks. To see how this works, let’s look at another poll. On May 20, the Wall Street Journal published an article headlined, “Critics of Bush on Terror See Caution Signs in Poll.” The story began:
[T]here are signs that [Democrats] need to be careful about pushing their critiques too far or too fast. In a new Wall Street Journal/NBC News poll, conducted during the weekend, Americans said by a 58% to 31% margin that they are satisfied with the measures the Bush administration took prior to the Sept. 11 terrorist attacks, based on the information it is known to have had available. … 58% of those polled said they think a full-scale investigation would be unproductive and too political.
Whoa again. Let’s look up the questions to which the Journal refers. The first one asked, “Are you satisfied or dissatisfied with the measures the Bush administration took based on the information they had prior to September eleventh?” The second asked, “Do you think that there should be a full-scale investigation into the handling of intelligence before September eleventh, or do you feel that this would be unproductive and too political?” At no point did the poll interviewers explain to respondents what “information” the administration had prior to Sept. 11 or how the intelligence was “handled.” The respondents were simply guessing, from what they knew so far, whether an investigation would turn out to be productive or unjustified. Arguably, at least, they couldn’t know how they would ultimately answer that question until they acquired more knowledge—which is what investigations are supposed to provide.
Ballot Box knows all about the fallibility of polls and predictions. A year and a half ago, he called Bush “toast” when Bush turned south in pre-election surveys. It was rash to declare Bush dead then. It’s equally rash to declare him invincible now.
