A couple of nights ago, during the nightly Daily Kos Elections Polling Wrap, the topic of discussion was whether or not the press was deliberately suppressing the breadth of the Obama post-DNC "bounce" in order to continue the appearance of a tossup race that could go either way. At the time I wrote that piece, I actively wondered if I was perhaps donning a tinfoil hat.
CNN's coverage of yesterday's trio of NBC/Marist polls seems only to underscore that point.
For those who missed it, the NBC/Marist polls found the president staked to 5-point leads among likely voters in Florida and Virginia, and a larger 7-point advantage in Ohio. With registered voters, the margins were spread out to a slightly larger spread of 7-9 points.
CNN's Gregory Wallace, writing up the polls on the media outlet's website, categorized Florida and Virginia as a tie. A tie.
This is a classic case of being technically correct and yet committing journalistic malpractice, nevertheless. Check out the characterization of those polls:
HEADLINE: Ohio trending for Obama; Florida, Virginia tiedThis is one of those classic examples of a journo knowing precisely jackshit about the concept of margin of error. Wallace, in this case, dubs the race a tie because a Romney lead could, in theory, fall within the margin of error. Which it could, if the actual result was at the absolute low end of Obama's MoE, and at the absolute upper end of the Romney MoE.
President Barack Obama has a narrow advantage over Republican presidential nominee Mitt Romney in Ohio according to a new poll released Thursday which also found the race for the White House locked up in Florida and Virginia.
In Thursday's polls, Ohio likely voters split 50% for Obama, 43% Romney, according NBC/Wall Street Journal poll, with 83% strongly supporting their candidate of choice.
Voters in both Florida and Virginia split 49% for Obama and 44% for Romney, a five point difference within the poll's sampling error of 3.1.
However, what Wallace conveniently ignores, presumably because he either did not know or did not care, is that the converse is equally plausible. Indeed, if you want to make the parsing argument that Wallace was not technically wrong to declare the race a "tie," you could therefore follow that it was equally accurate to write the header "Obama up double digits in Florida and Virginia". After all, with an MoE of 3.1 percent, an 11-point Obama lead in both states is just as plausible as an infinitesimal Romney advantage.
The Associated Press actually addresses how to deal with polling margins of error, and they are quite explicit about how to characterize races such as these (emphasis mine):
If the difference between the candidates is more than twice the sampling error margin, then the poll says one candidate is leading.Double the sampling error, in this case, is 6.2. By the AP standards, Obama is "slightly ahead," at worst. But he is closer to having an unambiguous lead than he is to being what the AP would characterize as a "close race" or "about even."
If the difference is less than the sampling error margin, the poll says that the race is close, that the candidates are "about even." (Do not use the term "statistical dead heat," which is inaccurate if there is any difference between the candidates; if the poll finds the candidates are tied, say they're tied.)
If the difference is at least equal to the sampling error but no more than twice the sampling error, then one candidate can be said to be "apparently leading" or "slightly ahead" in the race.
The bottom line is while there is a statistical chance that sampling error could account for the entirety of Obama's lead in the polls in Florida and Virginia, it is a very small chance, one that falls in just a small fraction of the range of possible outcomes. And to characterize the race as "tied," either in the headline or the text, falls damned close to willful misrepresentation of the state of the race.