Michael Barone

	JWR's Pundits World Editorial Cartoon Showcase Mallard Fillmore Mona Charen Linda Chavez Ann Coulter Greg Crosby Larry Elder Don Feder Suzanne Fields James Glassman Paul Greenberg Bob Greene Betsy Hart Nat Hentoff David Horowitz Marianne Jennings Michael Kelly Mort Kondracke Ch. Krauthammer Lawrence Kudlow Dr. Laura John Leo David Limbaugh Michelle Malkin Jackie Mason Chris Matthews Michael Medved MUGGER Kathleen Parker Wes Pruden Sam Schulman Amity Shlaes Roger Simon Tony Snow Thomas Sowell Cal Thomas Jonathan S. Tobin Ben Wattenberg George Will Bruce Williams Walter Williams Mort Zuckerman Consumer Reports

A turn in the polls?

A look at voters' party identification in polls conducted before and after September 11

http://www.NewsAndOpinion.com --
Since September 11, polls have shown almost unanimously positive job ratings for President George W. Bush and very highly positive job ratings for both Republicans and Democrats in Congress. They have also shown little or no change in the size of the constituencies of the two major parties, either in terms of party identification or in terms of which party's candidate people would prefer in elections for the House (this is known as the generic House vote).

Democrats have typically led Republicans by a few votes in both party identification and generic House vote. These are similar to the responses to these questions in the 1996-2000 period, which saw a virtual deadlock between the parties. Bill Clinton won 49 percent of the vote in 1996, and Al Gore and George W. Bush each won 48 percent in 2000. Republicans have led in the popular vote for the House by margins of approximately 49 percent to 48 percent in 1996, 1998, and 2000. Such stasis in electoral percentages has been unusual if not unprecedented in American history.

Now we have some evidence of a turn in the polls. Most of it comes from Ipsos-Reid, a nonpartisan firm, in polls cosponsored by the Cook Political Report. (Disclosure: Charles Cook, head of the Cook Political Report, is a collaborator in The Almanac of American Politics 2002, of which I am a coauthor.) Ipsos-Reid has been conducting national polls every weekend for nearly a year, and it took a look at voters' party identification in polls conducted before and after September 11.

In eight polls before September 11, 46 percent identified themselves as Democrats, and 37 percent as Republican. In 11 polls after September 11, 43 percent identified themselves as Democrats, and 42 percent as Republicans. A Democratic advantage of 9 percentage points became a statistically insignificant 1-point advantage. The results were similar in the firm's most recent poll on the question of which party they would like to see control Congress (a question Ipsos-Reid's Thomas Riehle thinks is a better indicator of electoral intentions than the generic House vote). The result there was 43 perce

nt Democrats, 44 percent Republicans. These results come in not long after two national polls showed statistically significant Republican leads in the generic House vote. The Gallup Poll conducted December 14-16 showed Republicans leading by 48 percent to 43 percent. The Fox News/Opinion Dynamics poll conducted December 12-13 showed Republicans ahead 40 percent to 29 percent (with an unusually high percentage undecided).

These results are out of line with most post-September 11 polls, which show the generic vote for House just about even or slightly Democratic. But one must treat these new polls with some caution. Any one of them may be the one poll out of 20 that is wrong.

(When you hear it said that the margin of error in a poll is plus or minus 4 percentage points, it may be left unspoken that the confidence level is 95 percent; that means that in 5 percent of polls-one out of 20-the results are more than 4 percentage points away from the results the poll would have gotten if it had surveyed the total population.)

And it should be noted that in the Ipsos-Reid polls the big change is the number of people willing to identify themselves as Republicans. This may simply be a matter of people who usually vote Republican suddenly becoming willing to wear the party label.

Moreover, a shift in party preference nationally may not be reflected in the outcome of House races, because only a small number of House races will be seriously contested-probably a number closer to 20 than to 40 of the 435 total. Personal factors or local issues could determine the outcome in many of these. And the high ratings for incumbents of both parties suggest it will take quite a strong shift in partisan sentiment to defeat members running for re-election.

So it is premature to conclude from the Ipsos-Reid, Gallup, and Fox numbers that a shift in partisan preference toward Republicans is underway. What these numbers tell us is to keep our eyes open to see if other evidence emerges. Political oddsmakers have long preached that the president's party almost always loses seats in the off-year election. But not quite always. In 1998 Bill Clinton's Democrats surprised just about everyone and gained a few seats. In 2002, if there is a change in partisan preference, George W. Bush's Republicans are likely to surprise just about everyone and gain a few seats-or even more, if the trend is enough to make winnable seats that currently seem safely Democratic.

Michael Baone Archives

JWR contributor Michael Barone is a columnist at U.S. News & World Report and the author of, most recently, "The New Americans." He also edits the biennial "Almanac of American Politics". Send your comments to him by clicking here.