Talk Elections

There is a new poll out from Muhlenberg College and Morning Call claiming Barack Obama’s job approval rating is down to a mere 35% in the state of Pennsylvania. in a generic match up, he does hold and edge on a Republican albeit with a ton of undecideds.
PRESIDENT – PENNSYLVANIA (Muhlenberg/Mcall)
Barack Obama (D-inc) 36%
Generic Republican 31%

http://www.mcall.com/news/breaking/mc-pa-obama-poll-20110819,0,1628054.story

Please stop posting clearly junk polls. You expect us to believe that over 30% of Americans are undecided?

Terry Madonna is one of the better pollsters out there, especially for PA.  I'd treat it Quinnapiac for PA.

This one is no where near a junk poll.  Madonna's poll have an exceptionally good record.

When neither candidate is even at 40%, it's junk.

Sample size: 380

Junk.

No, Madonna doesn't push the undecides, like Rasmussen.  All his early polls show this.


A 380 sample size is quite valid.

This one is not junk, and this is following Quinnipiac poll of two weeks ago.

http://blogs.mcall.com/penn_ave/2011/08/more-pennsylvanians-disapprove-of-obama.html

Obama's national numbers have been declining since then.

But look at other pollsters. They don't push leaners and still get over 40% for at least one candidate.

Replace Generic Republican with Perry, Romney or Bachmann and see how quickly Pennsylvania shifts to the Likely D category.

It is common in Madonna's polling, and he ends up with accurate results.

My mistake, Madonna isn't at Muhlenburg.  Not as great a poll as I thought.

I still wouldn't complain too much about the sample size, or the undecides.

Complete garbage.

Amusingly, the thread title doesn't even describe anything in this garbage poll (he has 36%, is up, and by 5%)

Hillary2012 has no other purpose but to post obscure junk polls.

Quinnipiac's sample size was 1,358 people.  Almost 1,000 more than Muhlenberg College.  380?  Come on.  That is a complete joke of a sample size for a state with over twelve million people.

Junk.

Also, the Quinnipac article you posted said he fares better than both political parties.  Everyone disapproves of everyone.

The sample size required for an accurate poll is independent of the total population size (assuming the population is fairly large)

1000 people gives the same MOE if the population is a 12 million or 300 million....

FYI

Can I get this brand in toilet paper?

Not a great poll, but it has a 5.5 MOE.  The Quinnipiac poll probably had a 3.5 MOE.  It is very possible that both polls are saying the thing, 40-43, possibly with a decline over the last fortnight.

Somebody has to do it.

Muhlenberg is only semi-interesting when they start their daily tracking poll a few weeks ahead of each election.

They said Casey would win by only 8 in 2006, but he won by 18. It seems they have gotten better over time though. They underestimated Obama by 5 points in 2008 and overestimated Toomey by 2 points in 2010. Maybe they nail it in 2012 ...

No offense, but you don't seem to understand how sample sizes work.

It never ceases to amuse me when people just don't understand the actual mathematical logic behind polling, and insist that valid statistical formulae don't apply because they don't "sound correct."

What I find strange is not that people don't know it (we can't all know everything about everything) but that people will comment so confidently on something they obviously don't know anything about. Furthermore, although this is a bit circular, someone who comments confidently on things they don't know must reasonably have been confronted with being wrong multiple times in the past. Yet, they still appear to be doing it.

This is the only thing of value in this thread.