PDA

View Full Version : Why did Cn/2 change how it gets voters to poll for its latest poll (showing 'tie'?)




sailingaway
08-22-2010, 10:42 AM
Did they just not like the results they were getting the other way, which came up to a 9.3% lead for Rand Paul last time over Jack Conway? (pretty consistent with Rasmussen's 9 point lead and Survey USA's 8 point lead?)

New poll showing 'tie'

" Sampling for this study was conducted using a Kentucky RDD sample. All interviews were conducted using a computer assisted telephone interviewing system. Statistical weights were designed from
the United States Census Bureau statistics."

http://politics.mycn2.com/wp-content/uploads/2010/08/Kentucky-Aug-19-poll.pdf

Prior cn/2 poll ( couple of weeks ago) showing Rand with a 9.3 % lead.

"Sampling for this study was conducted using a Kentucky Voter List supplemented with a new registrant voter list per our in house statistician's recommendation. All interviews were conducted using a computer assisted telephone interviewing system. Statistical weights were designed from the United States Census Bureau statistics."

http://politics.mycn2.com/wp-content/uploads/2010/08/7479-Kentucky-State-Tabs.pdf

Note this discussion of how the different methodology differently skews the results with the RDD approach newly used for cn/2s latest poll, 'inevitably' picking up people not registered to vote.

http://www.allacademic.com/meta/p_mla_apa_research_citation/1/1/6/2/8/p116287_index.html

TheDriver
08-22-2010, 10:51 AM
It fits their liberal agenda?

South Park Fan
08-22-2010, 10:52 AM
I was wondering how Rand could lose 10 points in a week. Thanks for pointing this out.

Lord Xar
08-23-2010, 01:07 AM
wouldn't be surprised if they really try to push this as a legitimate poll.. and kentucky is known for its less than honest elections.. I think the fix is in.

rich34
08-23-2010, 06:41 AM
We're gonna have to be on our toes come election day..

MRoCkEd
08-23-2010, 09:17 AM
Ed Morrissey:

The Kentucky race isn’t a slam-dunk for Republicans, but this poll isn’t a reliable indicator of the race, either. Its weighted sample has a 16-point advantage for Democrats, 54.2% to 37.5% for Republicans and only 8.2% independents. In 2008, in what most people would consider a banner year for Democrats, John McCain beat Barack Obama by a margin of 57/41 in Kentucky. If it takes a 30-point swing in the sample to get Jack Conway into a statistical dead heat with Rand Paul, that tells us all we need to know about the “opportunity” Conway has in November.