Reporting poll results: a double standard?

Jonathan wrote in SUSA Clinton Holds 16-Point Lead in Indiana
http://www.mydd.com/story/2008/4/14/1726 /17200

"The SUSA poll was in the field Friday through Sunday and had a margin of error of plus or minus 4.2 percentage points. As a result, the movement in the poll was not statistically significant."

While this is technically true, I will ask several questions and make several comments.

1.  The previous poll from IN (SUSA) was Clinton +9.  What would you be writing (and what would the Obamacrats be saying) if it went 7 points the other way to Clinton +2?

2.  I submit that all the Obamacrats here would be all over the "movement" in the polls.  Just recently, Liberal Avenger was talking* about "movement" within the Quinnipiac poll from PA which went from Clinton +9 to Clinton +6  *http://www.mydd.com/story/2008/4/8/12533 5/4902 Even though there was no movement in the poll, everyone was writing Clinton's obituary because the polls were "tightening" based off Quinnipiac, according to Liberal Avenger anyway.

3.  Since the movement is not "statistically significant," what does it matter if Clinton is now up 16 or 2 in Indiana?  It's all the same, right?  In your words Jonathan and those of statistical significance, they are both the same.  Yet, experience in the real world tells us they are not the same.  If you think they are, would you rather be up +16 from +9 or up +2 from +9?

In the future, I urge everyone to post with the same skepticism of the MOE as Jonathan did in his piece about Indiana.  Moreover, Clinton and Obama supporters should stop talking about "movement" unless the movement is 2x MOE, since that is what constitutes statistical significance.  

Accordingly, if in SUSA's newest weekly release on PA, Clinton is "down" to a 13 point lead from an 18 point lead a week ago, it should be reported that there was no movement, ditto if she is up 23 in their latest poll; the movement in either scenario would be within the MOE.

The reporting of polls is an important part of spin, so I doubt any of this will be absorbed in the collective conscience.  But, it's worth noting nevertheless.

Just think of how many times you've seen someone on either side here talk about "movement" when a poll changed from 45-44 to 47-44.  As educated people, you should know better than to spread such falsehoods.

Tags: Democratic primaries, Indiana, Pennsylvania, polls, statistical methodology (all tags)

Comments

6 Comments

Re: Reporting poll results: a double standard?

The over/under was Clinton +15 before the scurrilous "bittergate" Clinton/McCain/Media swarm.

Now it's back to Clinton +20, where it started a week or three ago.

Happy?

by Ddeele 2008-04-14 06:34PM | 0 recs
Re: Reporting poll results: a double standard?

Actually, I dislike Clinton and only like Obama better because he's in a better position to win in November IMO.  So your snark is not only unappreciated, it does not add to the serious discussion of poll results.

I encourage you to add more insightful comments to the community.  

Thanks for your comment.

by reggie44pride 2008-04-14 06:51PM | 0 recs
Reporting poll results: a double standard? YES!

I try to get the othe side out: New A.R.G. Pennsylvania Poll 45% to 45%

by Lefty Coaster 2008-04-14 06:57PM | 0 recs
a double standard? YES!

ARG is insane with their wild fluctuations.  There was a good explanation of why they are typically all over the place.  Have you read it?

I think if you average the last two ARG polls, you get a probable picture of the PA scene right now.  JMO

by reggie44pride 2008-04-14 07:04PM | 0 recs
Re: Reporting poll results: a double standard?

ARG POLL IS NOT 45-45. That was last week; this week they are at 57-37 Hillary. Next week they'll probably be at 52-40 Obama. None of these polls mean much as pollsters are all over the lot.

PA is easy: wait 7 days, you'll have the only poll that matters. When you get that poll, the others will adjust and you can have the same meaningless argument for IN/NC.

by STUBALL 2008-04-15 12:10AM | 0 recs
Re: Reporting poll results: a double standard?

ARG and PPP show wild fluctuations.  Other pollsters do not.  SUSA, for instance, rarely sees big jumps from poll to poll.  ARG finds 20 point jumps, commonly, which is bogus.

by reggie44pride 2008-04-15 12:57AM | 0 recs

Diaries

Advertise Blogads