WASHINGTON — Barack Obama beat John McCain in Tuesday night's debate in the eyes of undecided voters by 61 to 39 percent, according to a new online Ipsos/McClatchy poll.
Before the debate, the 389 undecided voters favored McCain by 55-45, but after it they shifted to favor Obama over McCain by 57-43 percent.
The online survey's value is like that of a large focus group; it is not a scientific random sample of the population, and so it has no statistical margin of error. Still, its results are illustrative of how many undecided voters perceived the debate.
While 57 percent said their opinion of each candidate wasn't changed much by the debate, 31 percent said it made them more favorable toward Obama, while only 18 percent said that about McCain.
McCain was judged more mean-spirited and disrespectful by 62 percent, to 38 percent who thought that about Obama. Fifty seven percent of the voters polled thought McCain demonstrated that he was tough enough to be president, while only 43 percent said that about Obama.
However, 65 percent said that Obama expressed his opinions more clearly, while only 35 percent thought that of McCain. Obama also was more likable, said 68 percent, versus 32 percent who said that of McCain.
Asked whom they trusted to handle the responsibilities of commander-in-chief, the voters polled favored McCain by 56-44 percent. He was favored 74-26 percent to defend the United States against foreign aggression, and by 61-39 percent to work with other countries to stop terrorism.
On the question of who'd better help the middle class achieve the American dream, Obama was favored by 73-27 percent, and by 63-37 percent on the question of who'd create more jobs. Such economic concerns are by far the dominant issues in American voters' mind at this point, recent surveys have found.
Obama also was favored 69-31 percent as to who'd better bring change to Washington, and by 58-42 percent on who'd get America on the right track.
How the poll was conducted. These are some of the findings of an Ipsos online poll conducted October 7-8. For this survey, a national sample of 389 undecided voters from Ipsos' U.S. online panel was interviewed online. Weighting then was employed to balance demographics and ensure that the sample's composition reflects that of the U.S. adult population according to Census data and to provide results intended to approximate the sample universe. Statistical margins of error are not applicable to online polls because they are based on samples drawn from opt-in online panels, not on random samples that mirror the population within a statistical probability ratio. All sample surveys and polls may be subject to other sources of error, including, but not limited to coverage error, and measurement error.
More from McClatchy: