( – promoted by Colorado Pols)
Denver Post reporters like to think of themselves as evidence-based folk. That’s what sets professional journalists apart from the rest of us, right?
So if you’re a clever headline writer at The Post, why would you put the following headline atop a story comparing a new Post Poll showing Romney with 48 percent of likely voters versus Obama’s 47 percent with the Post’s previous poll, five weeks ago, showing 47 percent for Obama and 46 for Romney? (Both polls had a four percent margin of error.)
“Romney Sways Voters!” [I added the exclamation point because it might as well have been there.]
Despite what this headline in the Sunday print edition said, the poll, released partially by The Post Friday with more details Sunday, showed a dead heat, with the difference well within the survey’s margin of error.
The fact that Romney gained two statistically insignificant percentage points since the previous Post poll is obviously meaningless.
Former Post columnist Mike Littwin got it right in two tweets today:’
D-Post fail: WaPo correctly calls its poll (Obama +3) “dead heat.” DP calls its poll (Romney +1) “shift in Romney’s fortunes.”
More on D-Post fail. Gallup: Romney +5. WaPo: Obama +3. Any poll fallible. Yet DP tells readers its 2-point swing significant.
If you’re wondering, a poll’s margin of error means that, if the poll were conducted 100 times, 95 times out of one hundred times the poll results would be within the margin of error, according to John Sides, an Associate Professor of Political Science at Georgetown University and a contributor to the Monkey Cage Blog.
So, in the case of The Post’s recent poll, 95 out of 100 times the results would show Obama’s percentage of the vote between 51 percent and 43 percent (four points up or down from his 47 percent) and Romney’s between 52 percent and 44 percent, Sides told me.
“We don’t have much confidence that Obama was leading in the first [Denver Post] poll or Romney was leading in the second–or that there’d been any change,” said Sides.