Frank Newport, Jeff Jones, and Lydia Saad
Frank Newport: Gallup Poll Daily tracking over the past week or so, along with a large sample USA Today/Gallup poll conducted last week with 1,600 national adults, indicates that the race for the presidency at this point is quite close. Barack Obama has held a modest lead for the most part, but over the past several days, even that small lead has evaporated to the point where Obama and John McCain are tied among registered voters. In fact, the tie has now persisted for two straight Gallup Poll Daily tracking reports (each report consisting of a three-day rolling average of more than 2,600 registered voters).
Indeed, Gallup's tracking results have been quite steady over the weeks -- spanning thousands of interviews -- showing little dramatic change from day to day or week to week.
A Time magazine poll conducted June 19-25 also shows the race close, with a slight 47% to 43% lead for Obama. A Fox News poll from June 17-18 shows a close race as well, with Obama at 45% and McCain at 41%.
Yet, two other polls -- Los Angeles Times/Bloomberg (June 19-23) and Newsweek (June 18-19) -- released over the past week show Obama with a double-digit lead over McCain.
It is more typical than not that well-done scientific polls measuring voter sentiment come up with roughly the same estimates. So the difference in estimates between these polls and the Gallup polls is unusual.
Specifically looking at the difference between the polls, estimates of Obama's share of the vote in our latest Gallup tracking from Monday through Wednesday is 44%. This compares to the 49% estimate for Obama in the Los Angeles Times/Bloomberg poll and the 51% estimate in the Newsweek poll (and to the 48% estimate among registered voters in our USA Today/Gallup poll conducted June 15-19).
Our present tracking estimate of McCain's current percentage of the vote is also 44%. The USA Today/Gallup poll from last week had it at 42%. But McCain was at 36% and 37% in the Newsweek and Los Angeles Times/Bloomberg polls.
The interviewing for these polls all span different time periods. If we look more specifically at the Los Angeles Times/Bloomberg poll, again, the interviewing period stretched from June 19 through June 23, and it found Obama at 49% and McCain at 37%. Our combined Gallup tracking for these same five days (involving more than 4,000 interviews with registered voters) shows Obama with 46% of the vote and McCain with 44%. So an apples to apples comparison (based on the same interviewing period) has a three-percentage point difference in estimate of Obama's share of vote, and a seven-point difference in estimate of McCain's share of the vote.
As is always the case, there are some slight differences in the way the polls are conducted. The Los Angeles Times/Bloomberg poll asks a "right direction/wrong direction" question before the ballot. Our Gallup Poll Daily tracking asks a registered voter screen before the ballot. The Los Angeles Times/Bloomberg poll includes the phrase "or would you vote for a candidate from some other party?" Our Gallup poll does not include this phrase. It is unclear how the order of these questions may affect the polling results.
Our extensive Gallup polling of the electorate simply did not pick up any major Obama surge over McCain in recent days. Broadly speaking, the two candidates remain closely matched among registered voters.
Jeff Jones: We've said before that things are shaping up as a potentially good year for the Democratic Party. That is evident in the party identification figures in our tracking polls. (The USA Today/Gallup polls show similar numbers). Over the past week of Gallup Poll Daily tracking, an average of 34% of Americans have identified themselves as Democrats, with just 24% saying they are Republicans. (About 40% do not identify with either major party.)
The Los Angeles Times/Bloomberg poll showed a slightly larger Democratic advantage, 39% to 22%, while the Newsweek poll had a similar 36% to 22% Democratic lead. It is often hard to rely on party identification as an explanation for poll differences, however, because it can vary depending on where it is placed in a given survey (i.e., what questions it follows; just as the ballot can) and can show short-term movement in response to the prevailing political situation. As a result, it is common for polls that show a party's candidate with an advantage on the ballot to also show that party doing well in terms of party identification. The two often move in sync.
But regardless of the precise numbers, all the polls generally agree that Democrats currently own a significant advantage in party identification among the American public. In fact, the advantage is the largest Gallup has recorded in recent decades. In all of its second quarter 2008 polling, an average of 35.7% of Americans have identified as Democrats, while 27.1% identified as Republicans. That 8.6-point party differential is the largest Gallup has found in any quarter since it made the shift to mostly telephone interviewing in 1988.
The Democratic advantage balloons to double digits when independents with a party leaning are included with the party identifiers. Using this definition of party affiliation, roughly half of Americans fall into the Democratic column as Democratic identifiers or Democratic-leaning independents (49% in Gallup Poll Daily tracking over the past week), while only 37% of Americans identify with or lean to the Republican Party.
From that perspective, McCain is overperforming compared with where he would be if Americans voted strictly according to their partisan inclinations, while Obama is not doing as well as he otherwise would be. That's due in large part to the fact that Republicans currently exhibit greater loyalty to their candidate (85% supported McCain in combined June 16-22 tracking data) than Democrats are to theirs (79% supported Obama).
Lydia Saad: There are invariably times during presidential election years when two or more polls provide conflicting pictures of where the race stands, and there is seldom a clear-cut explanation. That seems to be situation in this case.
"House effects" is an easy way out for discounting why one poll shows a more Republican or Democratic result than another does, but that only suffices when there are consistent differences between polls across time.
There may be something to this given observed differences between Gallup and the Los Angeles Times/Bloomberg poll. Looking back at all times since 2007 when Gallup and Los Angeles Times/Bloomberg have asked the Obama-McCain horse race question at similar times, the Los Angeles Times/Bloomberg poll consistently shows a wider lead for Obama -- ranging from 5 to 10 points. The Los Angeles Times/Bloomberg and Gallup tend to be close on the percentage for Obama, but the Los Angeles Times/Bloomberg typically shows a lower percentage than Gallup for McCain, and a correspondingly higher proportion of voters falling into Other/Unsure category.
Much has been made of the difference between the Los Angeles Times/Bloomberg latest 12-point lead for Obama versus the Gallup Poll tie reported on the same day, which is based on Gallup Poll Daily tracking from June 22-24. However, as noted previously, looking at Gallup Poll Daily tracking results for more comparable dates of interviewing (either June 19-21 or June 21-23), Gallup shows a two- to three-point lead for Obama on these days. Thus, the difference between the two polls is 9 to 10 points for the Obama lead -- similar to what we've seen at other times.
The story is quite different, however, relative to the Gallup versus Newsweek comparison. Until now, the two polls have shown remarkably comparable results for most similarly timed surveys since 2007. The only time the two were not within a few points of each other was a year ago in June, when Newsweek showed a larger lead for Obama than what Gallup found. It is difficult to compare the same periods of time in this instance because Gallup did not conduct its tracking interviews on one of the two days in which Newsweek was in the field.
One great value of Gallup Poll Daily tracking is that we complete the same number of interviews per night, and each night is an independent random sample. So along with our three-day rolling averages, we can look at the results by individual day to see what's happening.
Over the past week, the daily results have been quite stable with Obama and McCain closely matched in voter preferences. This is not a period of great volatility (as seen at other times) when sentiment is quickly shifting in favor of one candidate, or the lead is switching back and forth.
Given these stable trends across the thousands of interviews conducted as part of Gallup's tracking, it is surprising that other polling firms, particularly Newsweek, would be showing a double-digit lead for Obama at this time, but there is nothing in the Gallup Poll Daily tracking data to suggest why this would be the case.