"Logically Inconsistent" Responses

Last month, Gallup conducted an experiment that asked respondents about the war in Iraq. The responses, at first blush, appear to be "logically inconsistent." The questions were as follows:

            Do you think the United States can -- or cannot -- win the war in Iraq?

            Do you think the United States will -- or will not -- win the war in Iraq?

It should be immediately clear to even the casual observer that more people should say the United States can win the war in Iraq than say it will win. After all, if the United States will win, it first must be able to do so.

However, the results showed that only 52% of Americans said the United States can win, while a larger percentage, 58%, said it will win.

One reason for the apparent inconsistency is that the questions were administered to different people. A randomly selected half of the sample (which we called Form A) was asked whether the United States "can" win, while the other random half of the sample, Form B, was asked whether the United States "will" win.

As it turns out, the six-point difference between the two forms is not statistically significant, which means that though it looks like more people said the war "will" be won than said it "can" be won, the sample sizes are too small for us to be certain. Given the margin of error of each sample, about ±5 percentage points, the percentage of people saying "will" win could reasonably be no larger than the percentage saying "can" win.

Still, when the results are compared by political philosophy (liberal, moderate, and conservative), a statistically significant difference is found among conservatives, which raises again the problem of a logical inconsistency.

COMPARISON OF RESPONSES TO WHETHER THE UNITED STATES
CAN WIN OR WILL WIN THE WAR IN IRAQ

It is important to remember that there are two different groups being asked about the war in Iraq. If respondents from both groups were asked both questions, then presumably they would make the answers logically consistent with one another. But one group was asked only the "win" question, and the other group only the "can" question.

The results show that when conservatives were asked whether the United States "can" win the war, 64% said yes. Moderates and liberal were less optimistic, with only 49% and 36%, respectively, saying the United States can win.

When conservatives were asked whether the country "will" win the war, 81% give an affirmative response, even more than the percentage of conservatives who said the country "can" win. The percentages of moderates and liberals who said the country will win, however, were about the same as the percentages who said "can" win. If anything, the fact that about the same or slightly more of each group said "can" than said "will" win is just what we would expect from a logical point of view.

That 17-point difference between conservatives in Form A and conservatives in Form B is definitely statistically significant. That means that in the real world, or the larger population from which the samples were drawn, more conservatives said the country will win than said it can win -- a result that would be logically inconsistent if the sentiments came from just one person.

What do we make of these results? We know that conservatives are much more supportive of the war in Iraq than are either moderates or conservatives. In the latest CNN/USA Today/Gallup survey, for example, 70% of conservatives said the war in Iraq was not a mistake, while 52% of moderates and 67% of liberals said it was a mistake.

That stronger support for the war may cause conservatives to interpret the question about whether the United States "will" win in Iraq in a different way than moderates and liberals do. For these two groups, there is little difference in their assessments of whether the United States "can" or "will" win the war, suggesting that the "will" and the "can" questions stimulate similar feelings about the war. But conservatives appear to respond to the "will win" question as a clarion call to defend the war, as though the country's resolve is at issue, and they -- the conservatives -- will staunchly support that resolve. The "can win" question, on the other hand, appears to elicit a more subdued response from conservatives than the "will win" question, perhaps because conservatives see the "can win" question in more objective terms -- not as a challenge to the moral fabric of the country, but as a question that asks for a sober evaluation of the country's capabilities.

This explanation, of course, is quite speculative. We do not know for sure why conservatives respond with such vigor to the "will win" question, and with somewhat less vigor to the "can win" question. But I prefer my explanation to another alternative -- that conservatives are, as a group, just "logically inconsistent."