Tuesday, November 03, 2009

When We Talk about Media Bias...

This is the kinda thing we're talking about.

One year after he won a historical presidential election, a slight majority of Americans approve of the job Barack Obama's doing in the White House.

Fifty-four percent of people questioned in a CNN/Opinion Research Corporation survey released Tuesday approve of how Obama is handling his duties as president, with 45 percent saying they disapprove.

Is that really what the CNN poll shows? Well, Not really.
In an otherwise unremarkable poll of adults, as opposed to voters or likely voters, this result stands out, especially after the Obama administration’s attempts to both spin the numbers and blame George Bush for the economy. Fifty-four percent of respondents to the latest CNN poll disapprove of Barack Obama’s performance on the economy, a 17-point swing in six weeks. That isn’t the worst of the poll, either; 57% now disapprove of Obama’s performance on health care, a 19-point swing in that same time...

Another item missing from the poll and its report: the party split in the sample. The CNN/Opinion Research poll does not include any data on its survey respondents on party identification, a rather unusual omission for a political poll. Of course, that does allow CNN to skirt the issue of skewed samples, but it also makes the final report a lot less reliable, since we cannot test for sampling bias.

You'd think a 17-point swing in opinions in six weeks would be the lede. But, amazingly, that's not what CNN took away from its own poll. Of course, we have to wonder, yet again, if this were George W. Bush, what would the lede look like?

This story is a great example of how placement of information in a story (or leaving it out all together) affects the perceptions of readers. Anyone merely reading CNN's story wouldn't know that 57% of Americans disapprove of Obama's performance on healthcare, or that 54% disapprove of his performance on the economy. Why? Because CNN chose to focus on the vaguely worded "performance as president," versus his performance on specific issues.

Now, why would they do that?