Tip Jar. (4.00 / 1)

Editgrid.com is a great free resource for bloggers. Check it out!

---

by: NeuvoLiberal @ Mon Aug 27, 2007 at 20:59:08 PM UTC

I'm a sucker for graphs. Nice job. (4.00 / 1)

But I found it a little hard to resolve the discrepancies between your results and Chris's.

What polls did you include/not include? Can you tell why you're getting a different story than he is?

---

by: brklyngrl @ Mon Aug 27, 2007 at 22:57:56 PM UTC

[ Parent ]

Thanks. (4.00 / 1)

I haven't been keeping uptodate on Chris' posts on the 2008 horserace. I'd need to check his posts (if you'd like me look at specific ones, please link to them) before I can comment on his work.

However, in general, electoral college calculations are prone to carrying much greater margins of error, as the MOEs add up (nonlinearly) when you combine the results for the binary implications (that D would win state S over R) for several states. One needs to be very careful therefore when doing arithmetic from a given set of individual state polls.

If I find the time, I'll work out some hardcore math behind this and post it (I'm not a statistician, but have expertise in another quant. field of science). Thanks for your comment.

---

by: NeuvoLiberal @ Tue Aug 28, 2007 at 00:16:50 AM UTC

[ Parent ]

Ah. He did basically this exact same post today (4.00 / 1)

as you have here, with what seemed to me to be a similar methodology. Great minds think alike, I guess. :)

But he got totally different results, which is why I'm intrigued! That's the only reason I brought it up - I was flipping back and forth between the two trying to figure out what you used that he didn't, or vice versa. Here is a link to the post.  I'd love to hear your thoughts and his if you have time to comment (in it or here).

---

by: brklyngrl @ Tue Aug 28, 2007 at 00:42:17 AM UTC

[ Parent ]

He split the data by date, is only thing I could (4.00 / 1)

figure out. If you took all of the polls, that may be why.

---

by: brklyngrl @ Tue Aug 28, 2007 at 00:50:22 AM UTC

[ Parent ]

also, (0.00 / 0)

"What polls did you include/not include?"

Please find the polls used here: RCP 2008 national polls page. Thanks.

---

by: NeuvoLiberal @ Tue Aug 28, 2007 at 00:35:30 AM UTC

[ Parent ]

The reason I am different (4.00 / 2)

Is because I only looked at polls that included matchups for Edwards, Obama and Clinton in a single survey. Otherwise, you are comparing Honeycrisp apples to Red Delicious apples.

---

by: Chris Bowers @ Tue Aug 28, 2007 at 00:58:15 AM UTC

we could separate (0.00 / 0)

polls based on whether they included matchups for all three or not, and present them both separately and suitably/reasonably aggregated.

If time permits, i'll try to setup an elaborate spreadsheet (possibly a collaborative one) for tracking everything in one place in way that everyone can access the entire data easily and perhaps collaborate to build a useful resource.

There's too much partisanship and exaggerated projections in the poll reporting (i am not talking about your coverage or analysis here, just to make sure that you don't misread) by candidate supporting/promoting diarists  that we need a system where people get the facts and data unfiltered and in an unbiased format and sans the spins.

---

by: NeuvoLiberal @ Tue Aug 28, 2007 at 09:53:32 AM UTC

[ Parent ]

Added the tag (0.00 / 0)

'electability', since Chris is using it. I don't like the word because the word "unelectable" carries perceived negative suggestions, which people can spin away for as it suits them. I prefer "viability" (and degrees thereof) as that word tends to be more neutral and relatively spin-free.

---

by: NeuvoLiberal @ Tue Aug 28, 2007 at 10:09:16 AM UTC