As you remember, we've discussed some of the polls out there before from 'reputable' news sources.It's gotten to the point that outlets like Washington Post aren't even revealing their mechanics ( demographics and political affiliation of who was polled) anymore, so no one can tell exactly how skewed and biased the sampling is.
Here's CNN's headline: CNN Poll; Obama Holds 7 Point Lead Over Romney.
Sounds pretty daunting, no?
CNN/ORC don't give you their mechanics either, but they did leave one little clue in the PDF of the responses to individual questions that easily allows you to to extrapolate the oversample of Democrats...once you do the math.
The PDF shows they reportedly polled '1,010 adult Americans' with a margin of error of ' plus or minus 3.5 percentage points.'
Not registered voters, not likely voters, just 'adults'.
Nowhere does it give you any idea of the party affiliation of these adults...until you get to question number 23, which was limited to '419 Republicans or republican leaning independents' because it concerned Romney's possible VP pick.
So, let's do some math!
Obviously, if they asked only Republicans or independents leaning Republican about a question on a possible GOP running mate, it's because that's how many Republicans or leaners there were in the entire sample. No other explanation makes any sense.
So..419 amounts to about 41.5 % of the entire sample.
Subtract 419 from 1010 and you get 591 people who aren't Republicans or independents leaning Republican..or in other words, Democrats or independents leaning Democrat. They made up 58.5% of the sample, almost two thirds.
Subtract the Democrat percentage of the sampling from the Republican percentage and you come up with...a 17% oversampling of Democrats and Independents leaning Democrat.
Any wonder this 'poll' shows President Obama leading by 7 points?
Meanwhile Rasmussen, who uses likely voters show Romney up by 4 points, 47% to 43%, just to provide a second opinion on these matters.
The whole idea of these joke 'polls' like CNN's is the same as the notoriously skewed exit polls..to try and depress turnout and enthusiasm among Republican voters.
It's sad to see once reputable news organization bend over in this fashion, but it is what it is.
In reality, the race is pretty much even at this point with an edge to Romney based on President Obama's abysmal record in office.
People already know what they think of Barack Obama..now, they're deciding what they think of Mitt Romney, and his VP pick, hisconvention speech and his debate performance are going to seal the deal..or not.
It's his race to lose.
Rob: The PDF shows they reportedly polled '1,010 adult Americans' with a margin of error of ' plus or minus 3.5 percentage points.'
ReplyDeleteThat's right, but they only included registered voters in the results concerning the 7-point lead.
Rob: Subtract 419 from 1010 and you get 591 people who aren't Republicans or independents leaning Republican.
Only 911 registered voters, so 46% were leaning Republican, leaving 54% leaning Democratic OR non-leaning Independents.
The report should have shown the demographic breakdown. But it's important to realize that these are measured results, not weighted results.
Fox News Obama 9 points over Romney.
ReplyDeletehttp://www.foxnews.com/politics/2012/08/09/fox-news-poll-obama-lead-grows-as-romney-support-slips/
But of course Fox News is a famously liberal organization, right?
Hello Zachriel,
ReplyDeleteSince the PDF only names party affiliation in the question I noted, which was limited to Republicans and Republican leaning independents according to the pollsters themselves, it's a clear indication of how many were polled..as compared to Democrats and Democrat leaning independents.
But you raise two points of interest.
First, there's no indication of how many non-leaning independents there were. But if we look at the poll, 3% of both 'registered voters' and 'all respondents' chose neither candidate. That's a pretty good indication of how many 'non leaning independents' there were. It amounts to 30 respondents among all adults, and 27 if we're talking about registered voters.
Let's figure that into our results.
491 Republicans still equals 41.5% of the sample of all respondents. Subtract 30 respondents from 591, and you get 561 Democrats or democrat leaning independents. That works out to 55.6%, an oversampling of a mere 14.4%.
14.4%!!
The other point you raised is the smaller sample of 911 registered voters also provided.That is indeed the poll showing Obama with a 7% lead, and the one CNN used in its headline.
However, there's effectively no way to tell how many Republicans,Democrats, leaning independents, or non-leaning independents were polled as registered voters,because the pollsters (deliberately, in my opinion) don't include a breakdown. So I'm limiting my figures to the total amount of respondents,1010, because we can extrapolate those figures.
As you notice, Obama's purported lead is even greater among 'all adults', jumping to 11 points, not 7.
So anyway you slice it, this is a seriously skewed 'poll'.
Regards,
Rob
Anonymous,
ReplyDeleteA) This article is about the CNN poll, not FOX
B)If your point is 'even those neanderthal wing nuts at FOX say Obama's ahead', I would point out that FOX news is actually fairly middle of the road, and that a number of their most prominent commentators - Greta Van Sustern, Chris Wallace, Susan Estrich, Geraldo and of course, Juan WIlliams,among others - are Democrats. FOX just seems right wing because they have a few people who are admittedly conservative like Sean Hannity, and mostly because most of the rest of the media is skewed so far left.
BTW, looking at the FOX poll, they're quite open about having grossly oversampled Hispanics 'in order to analyze that subgroup'.
Other revealing figures from the poll...among interested voters who've been following things closely, Obama and Romney are tied, at 48%. That shows that the president's barrage of lying and misleading negative ads has merely kept things even.
And 15% of the people polled as Obama supporters say they're not sure and may change their minds.
Based on that, I still say it's Romney's race to lose, but we'll see in November.
An additional point. The polls, like CNN using registered voters are useless. Only 63% of those people voted in 2008.
ReplyDeleteRasmussen, which uses likely voters has Obama down by 4 points.