Monday, October 06, 2008

Polls and Media Bias

The first presidential election I voted in was in 1984, meaning I grew up in the heyday of the New Right, morning in America, the Christian Coalition and all that.

Perhaps no axiom was presented more on faith back then, and since, than the notion that the "media" (that monolithic entity) had a liberal bias.

Puhleeeze. Not sure it was true then. But now? Fox News is clearly pro-Republican (but only as a corrective! they shout, as though two wrongs make a professional argument), and everyone else is so afraid of the accusation of bias that they bend over backwards to not appear to be so. The problem with this is that it means distorting the truth--because sometimes the truth is not what the GOP wants to hear.

If you go to just about any site that gives you raw polling numbers (like this one)(or this one), you will see that Barack Obama has been making steady gains since the first debate and the economic crisis, is well ahead in most swing states, and--were elections held today--would win heartily. Polling shows he is ahead in 10 of 12 "swing" states, in several cases by double digits. The polling data was so disheartening to the McCain campaign that earlier this week they decided to pull out of Michigan, basically conceeding defeat in that state.

Yet, remarkably, as of today, nearly a week after the GOP has pulled out of that state, The Los Angeles Times has it listed as a "toss up" and the Washington Post still has Michigan "leaning" Obama, but by less than 10%.

Let me repeat that. The polls were so bad that the Republican party QUIT THE STATE, and a week later the LA Times says, "Eh, too close to call."

The Washington Post says that 174 electoral votes are from states "leaning" Republican and only 168 "leaning" Democrat, implying through the headlines not only that the race is close but that John McCain has a slight lead. The Times not only lists Florida and Ohio (where recent polls show Obama up by 8% points) as "toss ups" but also New Mexico, Oregon, New Jersey, and Pennsylvania. By contrast, Virginia, a state where the most recent polling shows Obama ahead by as few as 2 points (in Republican friendly polling) and as many as 10 (in some network polls) is not a "toss up" but squarely in the McCain column! CNN finally got around to putting Michigan in the "leaning Obama" category, and continues to list New Mexico as only "leaning Obama." Meanwhile, North Carolina, a state in which most Public Policy Polling data has shown Obama up by six points is listed as "leaning McCain" while Florida, which shows Obama ahead by a margin outside the margin of error and Ohio which currently shows Obama with a lead four times as great at the one McCain is alleged to have in the best NC polls are both shown as "toss ups."

I suppose one might argue that after the 2000 debacle where major media outlets called Florida for Gore based on exit polls that the media is reluctant to project a winner. But that's not quite the same thing as projecting a leader.

I really think the main reason is that the idea of a "close" election is a better story. It gives people reasons to tune in and drives ratings up. Who cares if it is true? Who cares if the media increasingly shapes the story by creating the impression that things are what they aren't and then wringing their hands about how strange it is that it is so close...creating a perception that Americans are more skeptical of Obama than the numbers say we are.

The second link above shows that even when taking the most McCain friendly polls (Rassmussen, aka Fox news), that show Virginia within the margin of error rather than a double digit Obama lead and Ohio being a tie rather than a 5-6 point Obama edge, that even the most McCain friendly polls show Obama winnign by about 130 electoral votes.

I'm reminded of the old 80s joke about how an American and a Soviet were in a marathon. The American won by 12 minutes. The next day Pravda reported: "In an international footrace, the Soviet runner finished second. The American runner was next to last."

The LA Times would probably just say, "The results were too close to call."


lauramorefield2000 said...

Interesting, Ken. Thanks for doing that research and sharing it.

I find myself having a bit of cognitive dissonance over this. The race still feels "close" to me, but the polls do say otherwise. I think your post helps explain the disconnect.

Kenneth R. Morefield said...

Welcome, Laura.

David Gergen suggested on CNN that some researchers fear that race can skew polling data--that is that some people will say they will vote for a Black person but then not do it or say they are "undecided" when they have no intention of ever voting for a Black person. Others on that network suggested that there was little evidence of that phenomenon during the Democratic primary (where the results mirrored the polls pretty well). Of course, one might worry that the sorts of people that Gergen worries about wouldn't be voting in the primaries.

Another reason the election might "feel closer" is that the leads are usually less than the percentage number of voters who are "undecided" in some swing states. I have my own feelings about this--I'm not sure how many "undecided" voters are really undecided and how many just like to wait until the last moment to declare.

Also, it's worth looking at polling that is just between McCain and Obama versus polling that includes McKinney, Nader, Barr, or other candidates. For instance, if one looks at Florida (, one sees Obama up between 2-4% points; when other candidates are mentioned, both Obama and McCain lose about five points, with a few mentioning other candidates and a larger percentage being "undecided" (8% in the Suffolk poll). This suggests to me that the higher % of undecided could be attributed to indifferent or ambivalent voters (don't like either one) as opposed to one side or the other having "soft" numbers. I've heard a lot of people say that Obama only has to "split" undecided voters with McCain, but it looks to me like McCain might actually be hurt a bit more by third party candidates because he NEEDS those votes, and because Obama's negatives aren't high enough to make people vote for McCain as "the lesser of two evils."

Finally, as we all know, there is math, and there is electoral math. The election can feel or be close nationally but break electorally. McCain's got double digit leads in Arkansas (not sure why CNN has that only "leaning" McCain), Mississippi, Alabama, and Tennessee. He gets 32 electoral votes for that, roughly the same amount Obama would get for sealing the deal in Michigan, flipping New Mexico (where he has a 5 to 6 point lead in a state Bush carried) and grinding out a win in Wisconsin (by no means assured but looking good. Stated more succinctly, McCain's margin of victory is liable to be bigger in the states he does carry, but Obama is likely to carry more states by a smaller margin. That suggests that things could still change with a major gaffe or some huge social change (I would look for Bush to announce some sort of victory or progress in the war in the week before the election), and I think the last thing the Obama camp wants is the perception of things looking good because a victory depends on high voter turnout.