Wednesday, March 31, 2010

Quinnipiac gets fishy.

It looks like my suspicions of the Party ID sample in the latest Ohio Quinnipiac poll were correct.

After poring through the numbers and some correspondence with the folks at Quinnipiac, I found that the Party ID numbers they used for today's poll were significantly different from that of their last poll.

February's poll broke out this way: 27% GOP, 30% Dem, and 37% Independent, with the rest answering with another Party or No Answer.

But today's poll broke out these Party ID numbers: 24% GOP, 33% Dem, 35% Independent, along with the rest being another Party or No Answer.

So Quinnipiac went from a 3% margin between Parties to 9%. On top of that, the Independents that had consistently been trending R, also shrunk by 2%.

Fortunately, this story has gone national of sorts, reaching the National Review's Campaign Spot.

Jim Geraghty points out an important data point to consider when looking at this new Party ID spread:
That 33/24 split among Democrats and Republicans really stands out, as the 2008 exit poll put the split in Ohio at 39 percent Democrat, 31 percent Republican. In other words, does Quinnipiac really think that the makeup of the electorate will be better for Democrats on Election Day than it was in 2008?
That's an incredibly important point to make.

Additionally, the two polls this year that publicly highlighted Party ID showed things to be far closer. PPP has it broken out as 42% Dem, 39% GOP, and 19% Indie/Other. The Ohio Poll actually had the Party ID broken out favoring Republicans, with approx. 49% GOP, 40% Dem, and 10% Independent.

Considering the political environment this year, an average of the two seems far more likely to be accurate than a Party ID sampling that shows a large Democrat advantage than the Year of Obama.

The fact is this:

Party ID when conducting a poll matters. The details matter. If the sampling doesn't accurately reflect the makeup of the electorate, you have a poll that illegitimately accentuates the numbers for one candidate or another.

Fortunately, this sampling was sniffed out. And it's actually surprising to hear since Quinnipiac has done a reasonably good job in the past of providing an accurate reflection of the electorate.

It just highlights how important it is not to take one poll as an accurate representation of the entire race. To do that, you need to consider an entire amalgamation of polls to gauge the overall trends.

We'll chat a little more about that tomorrow on 3BP....

1 comment:

  1. I don't understand how these supposed professional pollsters think they can get away with such ridiculous games like varying the sampling so wildly after just one month. Ridiculous. This is a perfect example of why I generally only trust Rasmussen.

    ReplyDelete

No profanity, keep it clean.

Note: Only a member of this blog may post a comment.