Commentary By Ron Beasley
The new Gallup poll indicates that Republicans Move Ahead in 2010 Vote for Congress
As Steve Soto points out there is a little problem with this poll.
What Gallup and the USAT aren�t telling you is that in the registered
voter sample they used in their late March polling, 44% of that
sample was unemployed, and that they used a weighted sample
with the same number of GOP/lean GOP voters as Democrat/lean Democrat
voters. I know unemployment is bad right now, but is it "44%" bad? And
at what point did Republicans catch the Democrats among registered
voters? I must have missed that.If you're going to build a poll based on a sample that is sitting at
home and 44% unemployed, and 47% GOP or GOP-leaning, presumably getting
all their information from Fox News telling them that it's all Obama's
fault, are you surprised at this result?
As Steve points out the other issues covered by Gallup show the results that one would expect from their flawed sample.
*The country has turned
against the health care reform bill;
*The GOP has now taken a generic
ballot advantage over the Democrats for the fall election;
*A majority feel
that healthcare costs and the federal budget deficit will get worse
as a result of the bill; and
*The public (incredibly) blames
the Democrats more for the vandalism and threats in the aftermath
of reform�s passage.
Where does this idea these polls have a less than random sample of registered voters come from?
ReplyDeleteAll the pages linked to claim a random sample was used. Weighting is never mentioned.
On healthcare the only screening mentioned is for adults and on the generic congress poll only (self reported) registered voters are said to have been screened in.
The pages linked to often mention different sample sizes, different screening and different dates strongly suggesting different cuts of, if not entirely different samples were used.
Maybe the 44% is not from bureau of labor statistics style "unemployment"? While 44% unemployment is high only 56% having paying jobs isn`t that low for the entire population. At least, not if you take into account kids, students, retirees, "homemakers", the hospitalized and disabled, volunteers, prisoners, off the books work AND record unemployment. IIRC the official US stats dont even include many who have been unemployed for a while discounting them as "discouraged workers".
Though the samples are explicitly said to only include adults or presumably people of voting age so child labor laws should not be part of the explanation. Quickly reading the results I didn`t see any 44% near anything about employment. Text searching wouldn`t work since the results are formatted in bitmap pictures. Is there maybe a PDF with raw results I missed? Did a link get lost somewhere?