Comments on: Top 10 Web tools for election season http://election.princeton.edu/2008/10/10/top-10-web-tools-for-election-season/ A first draft of electoral history Thu, 12 Nov 2009 02:35:25 -0500 http://wordpress.org/?v=2.8.6 hourly 1 By: Sam Wang http://election.princeton.edu/2008/10/10/top-10-web-tools-for-election-season/comment-page-1/#comment-1090 Sam Wang Tue, 14 Oct 2008 12:35:37 +0000 http://election.princeton.edu/?p=1821#comment-1090 I disagree - there are good reasons why FiveThirtyEight is popular. As I've said before, Silver has put together an extremely useful resource. In regard to fluctuation, one factor that works against him is overreliance on numerical simulation when an exact solution is possible. The problem can be repaired, but he will need to reach beyond his current toolkit. In regard to correcting for national polls, this seemed like a good idea for a while, and I was tempted to do it. But Silver himself pointed out that the post-RNC bounce for McCain/Palin was concentrated in frontier states. This would not have been predictable in advance by existing demographic models. This invalidates the use of national polls as a correction. A real consequence is that showing McCain pulling ahead in the EC was probably an error. I wrote about this <a href="/2008/09/12/state-poll-update-nearly-complete-and-a-surprise/" rel="nofollow">previously</a>. I disagree – there are good reasons why FiveThirtyEight is popular. As I’ve said before, Silver has put together an extremely useful resource.

In regard to fluctuation, one factor that works against him is overreliance on numerical simulation when an exact solution is possible. The problem can be repaired, but he will need to reach beyond his current toolkit.

In regard to correcting for national polls, this seemed like a good idea for a while, and I was tempted to do it. But Silver himself pointed out that the post-RNC bounce for McCain/Palin was concentrated in frontier states. This would not have been predictable in advance by existing demographic models. This invalidates the use of national polls as a correction. A real consequence is that showing McCain pulling ahead in the EC was probably an error. I wrote about this previously.

]]> By: Ethan http://election.princeton.edu/2008/10/10/top-10-web-tools-for-election-season/comment-page-1/#comment-1086 Ethan Tue, 14 Oct 2008 06:46:23 +0000 http://election.princeton.edu/?p=1821#comment-1086 I was hoping that PEC would make the list! I seriously think that there is no reason why 538.com could be gaining more attention than PEC. I mean, your meta analysis have proven record in the past being correct while Nate have proven record in predicting baseball. I specifically like your model because your model is somehow very less prone to fluctuations. When all national polls showed McCain leading, you still consistently showed Obama to have some EV lead although by a very tiny margin. 538.com at that time showed McCain leading Obama because Nate's model take into account the national polls too. I hope you are 100% correct again this time :) I was hoping that PEC would make the list!
I seriously think that there is no reason why 538.com could be gaining more attention than PEC. I mean, your meta analysis have proven record in the past being correct while Nate have proven record in predicting baseball.

I specifically like your model because your model is somehow very less prone to fluctuations. When all national polls showed McCain leading, you still consistently showed Obama to have some EV lead although by a very tiny margin. 538.com at that time showed McCain leading Obama because Nate’s model take into account the national polls too.

I hope you are 100% correct again this time :)

]]>