Anybody remember the recent thread that had great info about polls/moe/% lead and what it all means?

[elocs]

I thought that I had bookmarked it, but can't find it. I think it was very informative in understanding what the numbers really mean. For instance, let's say Obama has a 4% lead in a poll with a 4% margin of error so the tendency is to think the race is a dead heat when in actuality with that % lead and moe the odds that Obama would win are actually something like 98%.

This thread was early on a weekday and did not get too many responses, so I would love to see it posted again.

[fight4my3sons]

1. was it this..

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=132&topic_id=6469190

[elocs]

3. I'm not sure. I think the one I am looking for gave a detailed explanation

about the margin of error and the lead and how if the 2 numbers are the same does not mean the race is a dead heat as so many may believe. In other words, there may be many states where Obama does not have a big lead, and maybe even one within the margin of error, but the actual odds of his winning with that lead are very high.

[fight4my3sons]

5. here's the weekly edition

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=103x369809

Does that look more like it?

[elocs]

6. It looked similar, but I don't think it is the one. The poster really made the moe thing very clear

and I could have sworn the thread was made during the week and during the daytime, so it did not get many responses, but more than the half dozen or so here. Damn, why didn't I bookmark that? :crazy:

[babylonsister]

2. Maybe in Election Reform? Here's a link...

http://www.democraticunderground.com/discuss/duboard.php?az=show_topics&forum=203

[ms liberty]

4. If/when you find it, please post a link to this thread....

It sounds like something I would like to read. I've not been here as much recently, been really busy IRL.

[phrigndumass]

7. I couldn't find a DU thread, but I might have found the source for you

Did it look something like this?

http://www.fivethirtyeight.com/2008/07/tipping-point-v20.html



:hi:

[elocs]

8. I FOUND IT! HOORAY! Check the link for the table, but here's the gist of it:

Here it is: http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=132&topic_id=6451032

hokies4ever Donating Member (1000+ posts) Click to send private message to this author Click to view this author's profile Click to add this author to your buddy list Click to add this author to your Ignore list Thu Jul-03-08 11:58 AM
Original message
Understanding Election Polls 101

Alright, I'm tired of hearing people misinterpret polls, so here's a brief refresher in understanding how to properly interpret election polls. In particular, what do people mean by a 'statistical dead heat'? What's that margin of error really mean? If Obama is leading McCain 46 to 43 in a poll with a margin of error of 3%, is that a statistical dead heat? Well, here's a great answer:

The real question that one is essentially asking when interpreting the statistics of a poll is, "What's the percentage likelihood that Obama is leading McCain?" Or vice versa if McCain is ahead in the poll. All polls have random error in them (since you don't poll everybody, just a small sample size). Here is a useful table to tell you what's the percentage likelihood that one candidate is leading the other. The left axis is the margin of error stated in the poll (determined exclusively by the sample size). The top axis is how much one candidate is leading the other. Within the boxes is the answer to the question, "What's the percentage likelihood that Candidate A is leading Candidate B?"



The table comes from this website: http://www.washingtonmonthly.com/archives/individual/20...


Alright, now you can use this table to better understand a poll. So if you see a poll that has Obama leading McCain by 2 points with a margin of error of 3 points, that tells you that there's a 75% chance that Obama is truly ahead of McCain from those poll results.

One final point: The important variable here is the ratio of the percentage point lead vs. the margin of error. So there is a 75% chance that Obama is ahead of McCain if he leads by 2 points with a margin of error of 3 points. Also, there's the same 75% chance that Obama is ahead of McCain if he leads by 4 points with a margin of error of 6 points, since 2/3 = 4/6.

Hope this was helpful to people.

There is a very simple table in the middle that is most informative, so click on the original link to see it.