Hominid Views

There were four new polls released today. Quinnipiac University released large (> 1000 interviewees) polls in Florida, Ohio, and Pennsylvania. The Florida poll put Clinton ahead of her Republican challengers (Giuliani, McCain, Thompson, Romney and Huckabee) by at least 7%. It also shows Giuliani beating both Obama and Edwards. The pattern is identical in Ohio, except Clinton’s lead over Giuliani is only 4% and her lead over McCain is only 2%. In Pennsylvania, Clinton is tied with McCain and Giuliani, but she beats other opponents handily. Giuliani is tied with Obama and beats Edwards by 3%.

The other new poll is from Rasmussen in Arkansas where Clinton beats Giuliani and Romney but loses to Huckabee. Obama beats Giuliani but loses to Huckabee. Huckabee’s strong showing is hardly surprising—he was a very popular governor of Arkansas from 1996 to 2007.

Today I’ll focus on Clinton–McCain state head-to-head polls. Since my last analysis ten Days ago, McCain has slipped a little. He previously had a 32% probability of beating Clinton in a general election. If the election were held today, however, his chance of beating Clinton has shrunk to 21.6%.

Compared to the previous analysis, the biggest losses occur in Florida, where McCain’s probability of taking the state has shrunk from 94% to 28%, Iowa, where his chance of taking the state decreases from 84% to 19%, and New Mexico where his chances decline from 98% to 73%. To balance those loses, McCain gains in Pennsylvania where his chances go from 3% to 50% and Massachusettes, where he goes from a probability of 0% to 13%.

This result is based on 10,000 simulated elections using polls conducted in the last month whenever possible. Clinton won 7,719 times, McCain won 2,157 times, and there were 124 ties.

The trajectory in this race can be seen in the following figure, based on weekly simulated elections using polls taken in the preceding month (whenever possible).

(Click here to see the sequence animated over time. Details of the methods are given here.)

Even though McCain has slipped a bit in the last week, he continues to be more competitive than the other Republican candidates in a general election match-up. It is interesting, then, that McCain performs miserably against other Republicans in the Iowa caucus polls, the New Hampshire primary polls, the Nevada caucus polls, the South Carolina primary polls, the Florida primary polls, and the Michigan primary polls.

A comparison of state head-to-head polls shows that in each of these primary states (except SC where there are no data), McCain performs as well as or better than Giuliani against Clinton. This table shows the normalized differences between pairs of candidates (Republican poll result minus Clinton poll result) using pooled polls from the last month (or the newest single poll if no poll was taken in the last month):

On average McCain is down 1.2% to Giuliani’s 3.7% in these state head-to-head polls.

In national primary polls (here or here), McCain doesn’t do particularly well—he is in a four-way tie with Huckabee, Thompson, and Romney, way below Giuliani. Go figure.

It seems just a little illogical to me that the party nominees are selected more on the basis of a popularity contest within the party than they are based on measured performance against potential challengers. But there are many things that are not sensible about the U.S. presidential nomination and election system.

Here are the results of the current time period in the Clinton–Giuliani match-up.

• 10000 simulations: Clinton wins 77.2%, McCain wins 21.6%.
• Average ( SE) EC votes for Clinton: 286.1 ( 21.6)
• Average (SE) EC votes for McCain: 251.9 ( 21.6)
• Median (95% CI) EC votes for Clinton: 286 (245, 327)
• Median (95% CI) EC votes for McCain: 252 (211, 293)

^* denotes that an older poll was used

Details of the simulation methods are given here.