« Woo! | Main | 1/24 ARG Daily »
January 24, 2004
Exclusive!
Welcome from Kaus Files.
This is my own private New Hampshire tracking poll (see below for methodology):
| Projected | 1/23 | 22 | 21 | 20 | 19 | 18 | 17 | 16 | 15 | |
| 46 | Kerry | 39 | 32 | 30 | 31 | 21 | 19 | 20 | 19 | 17 |
| 23 | Edwards | 16 | 12 | 10 | 10 | 8 | 8 | 8 | 8 | 8 |
| 11 | Clark | 16 | 22 | 20 | 17 | 19 | 19 | 19 | 21 | 20 |
| 14 | Dean | 13 | 15 | 18 | 22 | 27 | 28 | 28 | 28 | 28 |
| 6 | Lieberman | 6 | 6 | 7 | 7 | 7 | 7 | 6 | 7 | 7 |
| Other | 10 | 13 | 15 | 13 | 18 | 19 | 19 | 18 | 20 |
Methodology: ARG produces a 3 day average tracking poll every night. For example, they do a nightly poll on Monday, Tuesday and Wednesday. Then they take an average of each candidate's support on those 3 nights, and that is the number for Wednesday. It's a good method, but it delays the results of a bounce or surge, and since there are only 3 days to go until the New Hampshire primary, that surge or bounce is what I'm interested in right now.
So, I set up the 3 day averages in a spreadsheet and doing some equations and educated guesses, I've figured out what their nightly poll was telling them each day. So, tomorrow, when they release the next 3 day tracking number, I already know the previous two days' daily poll numbers and I can simply work backwards from the average to report what each candidate is actually polling that day.
Here's an example of how/why it works: The day after Iowa (Jan 20th), Dean had a 2 point lead against Kerry in the ARG tracking poll. My data, however, shows that on that day Kerry was actually outpolling him by 9 points. That nine point lead didn't show up in the 3 day poll for 2 days, and by that time, my data showed Kerry opening up a 17 point lead, which is now the conventional wisdom.
My 'poll' data accurately showed Kerry's huge surge days before it was the CW, and now it's showing Edwards' sharp climb, Clark's sharp decline, and Dean's bottoming out. Will it be the new CW come Wednesday morning? I think so, but we'll find out. My projection for the final results are based on my own momentum tracking. I'll update these everyday too. Hope you'll be back to check them.
One final note: The "stock chart" in the left column won't be that accurate until states actually start voting and we get some more solid close-to-election polling. It still reflects the pre-Iowa CW, but since most states haven't been polled since then, I don't have much of a choice. But it does show momentum. A candidate needs a "50" to win the nomination. It reflects Dean's early lead in money and his ability to garner 25% or so in many states because of his name ID. Clark's 2nd place showing in the poll reflects his existence as the anti-Dean. As Dean becomes less likely, so does the need for an anti-Dean. Once Kerry or Edwards or anyone else starts actually winning primaries, they'll rise and Dean and Clark will fall. As soon as Lieberman drops out, he'll go down to 0. So stay tuned there too.
Posted by Chris at January 24, 2004 01:23 AM
Comments
Much props on the major link via the Kaus Files.
Posted by: Matt at January 24, 2004 10:58 PM
Thanks buddy.
Posted by: Chris at January 24, 2004 11:05 PM
He's right, the old photo was better. What's with the cutout job here?
Posted by: Wes at January 24, 2004 11:11 PM
It's my new drivers license photo. Maybe I'll replace it after the Kf stuff dies down with a better one.
I thought the old photo was a little too Matthew Yglesias.
Posted by: Chris at January 24, 2004 11:17 PM
Is there any chance that you could post that spreadsheet? I am curious what the educated guesses you make are. Doing this kind of thing with stats can be very misleading. Even if you had the data you are dealing with a sample size of about 100, leading to a huge margin of error. When you add in some error from "educated guesses" it can get very difficult.
But mostly I am just curious how you are doing this, rather than wanting tom impose some kind of intellectual purity on your reporting of the numbers.
Posted by: Rich at January 25, 2004 01:49 PM
Post a comment
Thanks for signing in, . Now you can comment. (sign out)
(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)