Last week, an Ipsos Public Affairs poll reportedly showed that, by a margin of 50% to 44%, Americans want Congress to impeach President Bush if he "did not tell the truth about his reasons for going to war with Iraq." The poll was commissioned by AfterDowningStreet.org, self-described as "a grassroots coalition that supports a Congressional investigation of President Bush's decision to invade Iraq in 2003."
Bob Fertik, co-founder of the Web site, expressed his astonishment at the poll findings ("The results of this poll are truly astonishing."), noting that other polls had previously shown widespread public opposition to Bush and the war in Iraq, but this poll gave the "other half of the story" -- "that a solid plurality of Americans want Congress to consider removing Bush from the White House."
But the poll question didn't ask whether Congress should impeach Bush. It asked if Congress should impeach Bush if he lied. While Fertik obviously believes that Bush lied, many people do not. The last time Gallup specifically asked if people believed the "Bush administration deliberately misled the American public about whether Iraq has weapons of mass destruction" was in a July 22-24 poll, and at that time, just 51% said yes, while 47% said no.
If we take the results of the Ipsos poll literally (and there is reason not to do so, as discussed later), that would mean that only half of the 51% who think Bush lied would want the president impeached, or about a quarter of Americans, rather than a "solid plurality" as Fertik claims.
But the Ipsos poll did not ask whether people thought Bush lied, and instead just asked the one hypothetical question, allowing the sponsor of the poll to post misleading interpretations of the results on the Web site.
There are at least two issues to address in this example. First, how do we interpret the hypothetical question about impeachment, when the hypothetical situation itself is never explored? And why would a reputable polling organization allow its data to be used in this clearly biased way?
Interpreting the Hypothetical Impeachment Question
On the AAPORNET Web site, restricted to members of the American Association for Public Opinion Research (AAPOR), one writer expressed disbelief that so few people would want to impeach Bush "if" he had lied about the reasons for going to war. Obviously, that writer assumed that respondents had taken the hypothetical situation literally, but there is good reason to believe that people read into the question being asked. In this case, I believe that many people said they would not impeach Bush "if he lied," because they believe he did not lie. Thus, they were not going to say "yes" because they did not want their responses to be interpreted as support for impeachment -- which is exactly how Fertik interpreted the affirmative responses.
The release on AfterDowningStreet.org mentions a "passion for impeachment," but if one wants to test whether such passion really exists, there are some objective ways for doing that.
Perhaps the easiest question would be simply, "Do you think the Congress should vote to impeach President Bush, or not?" We shouldn't provide any reasons for the impeachment, because to do would bias the question in favor of impeachment.
It is possible that even in the absence of any effort by members of Congress to pursue impeachment, about a quarter of Americans would still say yes to the impeachment question. In fact, it could be that for any president at almost any time, there would be a significant percentage (though much less than a majority) of partisan opponents who would support impeachment when asked in a poll.
Gallup has no trend on an impeachment question asked apart from times of controversy. We know that most people opposed impeachment of President Richard Nixon until very late in the Watergate investigation, and that a majority of people opposed the impeachment of President Bill Clinton even as he was being impeached. So, it is unlikely that a majority or even a plurality of Americans would support impeachment of Bush at this time, given the lack of any congressional effort -- by either Democrats or Republicans -- to introduce articles of impeachment.
Still, it is possible to design some objective questions to measure impeachment sentiment among the general public. Asking about impeachment if Bush lied, and then not exploring the extent to which people believe Bush lied, is clearly not the way to discover what Americans really think about the issue. This tactic is so clearly biased, it raises the question of why Ipsos participated in the effort at all.
Polling for Interest Groups
The AfterDowningStreet.org Web site indicated that many activists, led by Democrats.com, urged all of the major polling organizations to include an impeachment question. But when none did, the organization funded the one question on the Ipsos poll. Apparently, the organization intends to continue funding such questions in the future.
Thus, it appears that Ipsos granted the AfterDowningStreet.org request, because the organization provided the money to include the question on a poll. Ipsos can make the case that its methodology was sound and the question wording itself is not biased. And Ipsos could argue that it has no control over the use to which these objective results might be put.
But that would be an abrogation of responsibility. Ipsos' reputation is being used to support a highly biased interpretation of data. And Ipsos cannot simply wash its hands of the issue. The fact is that a single question of the type that Ipsos included cannot provide a meaningful measure of public sentiment toward impeachment, and for its own reputation, Ipsos should officially disown the interpretation on its own Web site.
In the future, all polling organizations who wish to protect their reputations should either not include questions for special interest organizations that have such a clear ax to grind, or they should insist on a set of questions that can objectively measure public opinion on the issue in question. They should also require that the sponsoring organization clear any public release of the data with the polling organization, in order to ensure that the results are being accurately portrayed.
If polling organizations do not protect their own reputations, the whole polling industry suffers.