Polling

[cawacko]

How does a poll come out one day saying Obama has a 15 point and then the next day a poll shows they are tied when nothing big happened? I'm not asking if Obama should be leading or tied just curious how two polls get such disparate results?

 

[Cancel 2016.2]

How does a poll come out one day saying Obama has a 15 point and then the next day a poll shows they are tied when nothing big happened? I'm not asking if Obama should be leading or tied just curious how two polls get such disparate results?

1) Sample size

2) The manner in which the questions are asked

3) The wording of the questions

4) Margin of error.... this is not as likely, but theoretically if you have a MOE of say 4% and "real" number is Obama up 4... then you could potentially have two surveys where one came in at a tie and the other had Obama up 8%. But no way can this be the cause of a 15% swing on no news.

 

[cawacko]

1) Sample size

2) The manner in which the questions are asked

3) The wording of the questions

4) Margin of error.... this is not as likely, but theoretically if you have a MOE of say 4% and "real" number is Obama up 4... then you could potentially have two surveys where one came in at a tie and the other had Obama up 8%. But no way can this be the cause of a 15% swing on no news.

The L.A. Times poll had Obama up 15 points yesterday and Gallup claims its tied today. Just seems wierd.

 

[Cancel 2016.2]

The L.A. Times poll had Obama up 15 points yesterday and Gallup claims its tied today. Just seems wierd.

the Times/Bloomberg poll had him up 12. Gallup even. Rassmussen Obama up 4. Newsweek had Obama up 15.

http://www.realclearpolitics.com/epolls/2008/president/us/general_election_mccain_vs_obama-225.html

Shows the sample sizes and links to each of the polls. You might find your answer there.

Gallup and Rassmussen use larger sample sizes, but the Times poll is large enough that the margin of error still wouldn't explain such a difference. Might be the wording or the manner in which it is asked.

 

[Epicurus]

Like SF said, different sample sizes and demographic assumptions on the part of the pollster.

The most obvious would be if the question was worded differently.