The difference between the two PA polls

[BzaDem]

The CNN poll (O+12) was conducted from Oct 23-28. It looks like the Mason-Dixon poll was conducted from Oct 27-28. Something tightening happened within the past few days, and it appears to have affected PA (though according to Nate Silver, Mason Dixon has a slight Republican lean, so the real margin is most likely higher than 4.

[Thrill]

1. It didn't tighten that much

[Jennicut]

2. No way it tightened that much. That poll is really crappy as are all the Mason-Dixon polls.

Also, people need to remember to take the average of all the polls together. The averages at RCP were nearly right on for the 2004 election. You cannot cherry pick polls that are really favorable or unfavorable. It simply does not work that way.

[Kahuna]

3. Do you have a link to the MD Oct 27-28 poll? nt

[BzaDem]

16. Here's the only link I have:

http://thepage.time.com/2008/10/30/pennsylvania-hope-for-mccain/

[Kahuna]

17. Thanks!

:hi:

[Schulzz]

4. When you look at all PA polls, McCain's support is always at 41-43%

Only Rasmussen has McCain's support higher, but they have only 1% undecideds left, so McCain can't close the gap.

http://www.realclearpolitics.com/epolls/2008/president/pa/pennsylvania_mccain_vs_obama-244.html

[nickn777]

5. even though M-D polls are crap there is one constant...

notice McCain's number at 43 even in M-D...

There is no surge to McCain. He is stuck at 43 in Pennsylvania. He is stuck at 43 in most battleground states. He is stuck at 43 in the RCP Avg.

[Zynx]

6. You are not a particularly good analyst of polls. Sorry. You're just not very good at this.

You are comparing two completely different polling firms, disregarding all other polls that show the same thing as CNN that have been taken in a more contemporary fashion, and trying to infer a trend of tightening that probably does not exist.

[BzaDem]

15. I admitted there were two completely different polling firms.

As far as tightening, there has been tightening (outside the margin of error) over the last several days in Zogby, Gallup, Rasmussen, and R2K.

If you have any PA polls taken in a more contemporary fashion (i.e. last two days only), please share. I might have missed some.

[MadBadger]

7. Mason-Dixon proved what an awful pollster they were in the primaries. For example, Georgia

I believe their final poll had Obama up only 47-41. I believe he won by over 30 points. They had Clinton winning 47-41 in Missouri,

[fujiyama]

8. My guess is

they use conservative polling models, not in an idealogical sense, but one which generally doesn't include "less likely" voters like minorities.

Selzer, based out of IA, is saying that many of the polls are ignoring the extent of black turnout. And she may be right, as evidenced by black early voting turnout especially in the south.

I think pollsters are having a hell of a time figuring out an accurate model this time. Especially considering the rise in registered voters and the enthusiasm and interest this race has generated.

[Mass]

9. 2 different polls by 2 different pollsters, for a start. Also, the MD poll from today shows an

improvement compared to the previous on.

And, despite all this noise, Pennsylvania is still +4.

[Mass]

10. This poll from the Marist was done on 26-27 and has Obama up by 14 among LV

Short of a cataclysm happening the 27 and 28, I would stop worrying about this poll.

http://www.maristpoll.marist.edu/PApolls/PA081029.htm

[GOPNotForMe]

11. It's one poll!

It's an island of red in a sea of blue. EVERY other PA poll for WEEKS, including the daily tracking poll, has shown a huge margin for Obama. If multiple polls start coming out showing the race tightening in PA, then we'll have a trend on our hands. Until then, as far as I'm concerned, this M-D poll is an outlier.

[Joe the Revelator]

12. Look at Mason Dixon's last VA poll

Their model is making the election closer across the board.

[faithfulcitizen]

13. The "socialist" drumbeat may be having an effect. So says the obamacans in family.

:shrug:

[aaaaaa5a]

14. I don't understand...

If there is a tightening in PA over the last 3 days, why does the PA Morning tracker which is good through 10/29 show Obama increasing his lead?


I think the MD poll is an outlier. Let's face it, PA is being polled alot. If you poll a place 10 times a week, you are bound to get one result that is a little off.

[Bleacher Creature]

18. There's way too much to be suspect about the M-D poll

First, with all of the time and money both candidates have spent in PA, do you really believe that 10% are undecided? Also, they've only polled PA twice now, and both cases were extraordinarily close. Finally, the number doesn't square with ANY other poll.

If it looks like an outlier and walks like an outlier, it probably is an outlier.