Combining Polls

The proliferation of polls makes modern presidential campaigns more fun to watch than those of earlier periods. There are so many polls that it's not hard to figure out where a race stands at a given point, even if it's very unclear where it is headed.

Looking at an average of poll results is always more prudent than focusing on just one survey -- particularly one that shows something substantially different from previous polls. Even if a given poll's overall finding is dead-on, its sample size is probably too small to put much credence in the results for various demographic and political subgroups.

That's why it's fun to look at combined surveys -- taking two identically worded and conducted polls, then aggregating the results so that the sample size of the subgroup is doubled and provides a more reliable snapshot of that subgroup's views.

Combining the two most recent Cook Political Report/RT Strategies polls -- those conducted February 15-18 and March 29-April 1 -- yields results based on 1,642 interviews with a margin of error of only plus-or-minus 2.4 percent.

This total included 722 registered Democrats and independents who lean Democratic (a combined subgroup with a 3.7 percent margin of error), and 590 registered Republicans and independents who lean Republican (margin of error: 4.0 percent).

In the race for the Democratic presidential nomination, Sen. Hillary Rodham Clinton of New York led the pack with 41 percent, followed by Sen. Barack Obama of Illinois with 19 percent; former Sen. John Edwards of North Carolina, 17 percent; and New Mexico Gov. Bill Richardson, 4 percent. Everyone else had 2 percent or less.

Clinton's 18-percentage-point lead among African-Americans was only slightly smaller than her overall lead of 22 percentage points: She drew the support of 43 percent of African-American respondents, compared with 25 percent for Obama and 7 percent for Edwards.

Clinton's lead among women was 25 points -- 45 percent versus 20 percent for Obama and 17 percent for Edwards. Among men, her lead was 20 points -- 37 percent versus 17 percent each for Obama and Edwards. While Clinton drew 42 to 44 percent support in each of three age groups, she dipped to 38 percent among those ages 35 to 49 -- the weakest age demographic for her party as well.

Obama scored best (23 percent) with the youngest voters, those 18 to 34, and worst (12 percent) among those 65 or older. Edwards's support didn't vary much among age groups. The support by education for Clinton was the exact opposite of that for Obama: Clinton scored best (49 percent) among those with a high school education or less, in the middle (42 percent) among those with some college, and worst (32 percent) among those with college degrees.

Obama performed best (29 percent, only 3 points behind Clinton) among those holding college degrees, in the middle (15 percent) among those with some college, and worst (13 percent) among those with a high school education or less. Clinton did best in the Northeast, with 50 percent, and worst in the West, 34 percent. Obama and Edwards both did best in the West with 23 percent and 22 percent, respectively, followed by the Midwest, with 19 percent and 18 percent.

On the Republican side, former New York City Mayor Rudy Giuliani led the field by 22 points when Newt Gingrich and Fred Thompson were excluded. He polled 42 percent, followed by Sen. John McCain of Arizona at 20 percent, former Massachusetts Gov. Mitt Romney with 9 percent, and everyone else with 3 percent or less.

There was much less variation in candidate support among the various GOP demographic and political subgroups than there was among the Democratic ones. However, McCain's support was clearly linked to higher education: He did best among college graduates. And McCain was stronger in the West than in other regions; Giuliani was weaker in the West than elsewhere.

Although I remain very doubtful that Giuliani can win his party's nomination, his support is uniform across most demographic groups. If he can maintain his popularity into the fall, when the public's scrutiny of the candidates will intensify, his skeptics will be forced to recalculate his chances.