Wednesday, April 30, 2008

Obama's Slide: Is Clinton Taking Advantage?

One of the biggest complaints about these maps that I've been producing for over a month now (see right side bar for the links) is that they pull in polling data from way back in February and give each of those polls (recent and long ago) equal weight. And that's a legitimate argument. Clinton supporters (and the detractors of my methodology) have chimed in in the comments sections of some of these posts to argue that Obama's good standing in the polls, post-Super Tuesday and pre-Jeremiah Wright, is propping up his averages, giving the appearance of competitiveness against McCain. My argument for including all the polls from Super Tuesday on has been that that was the point at which this became an even, two-person race for the Democratic nomination. The argument that we should expect some decay in a poll's value as it gets older, though, is still a valid one. I would counter that it is still necessary to include all the data from Super Tuesday forward, but to discount the early polls and weight the more recent ones.

Since there is a range in the number of polls per state
(a low of one and a high of 15) there was some variation in the number of polls that were considered recent or dated. The following rules have been adopted to deal with these differences:
  • If there has only been one poll conducted in a state, that is all the information on that state. Those poll numbers are left as they are, neither discounted nor weighted.
  • If a state has had two or three polls since Super Tuesday the more recent one was given extra weight while the dated one (or two were) was discounted.
  • Any state that had four polls over this period had the two most recent polls weighted while the earliest two polls were discounted.
  • Finally, in all the states that had five or more polls the three most recent polls were given added significance while all the other polls were discounted to account for decay over time.
The big question: What effect does this have on the electoral college projection maps? If the recent negativity around Obama's campaign has any relevance, there should be a noticeable shift in his numbers against McCain in the hypothetical general election polls pitting the two against each other. In other words, adding in a decay function and giving more significance to the newer polls should be expected to hurt Obama. Relatedly though, is Clinton taking advantage of her opponent's slide? She has picked up some points on Obama both nationally and in the states yet to hold nominating contests, but is that translating to a possible battle against McCain? Is she able to make a better case that she would be more electable against McCain than would Obama?

To the maps!
In comparing the weighted Obama map to the unweighted one (see link under Recent Posts at the bottom), several things are clear. There is a lot less light blue and purple and much more brown. Many of the states that were leaning toward Obama or were toss ups favoring him moved into the McCain column. Whereas last week's maps showed an even split in the number of toss up states between Obama and McCain, this map has 11 of the 15 total toss ups favoring McCain. As a result, the race between the two in the electoral college has gone from a virtual tie to an 80 electoral vote advantage for McCain (309 to 229) with 174 toss up electoral votes. Among the states where the candidates were either strongly ahead or held a solid lead, Obama led 192 electoral votes to 172. McCain, then, took 137 or those 174 toss up electoral votes.
And Clinton? Her map and the resulting electoral college projections are nearly identical to the unweighted map. There are changes in the map on the margins, but the electoral vote tally is exactly the same (304 to 234 for McCain). That indicates that she has managed to maintain a certain standing against McCain, but that Obama's recent troubles have not changed the perceptions of those polled in regard to her standing against the Arizona senator.
And what of the difference each candidate makes in the various states? Here too, the maps are largely similar. The states shaded on each candidate's map are the same with the weighted data as they were before the transformations to the data were made. The difference is the magnitude of the differences. The Obama map is much lighter now. There are many more yellows and greens now than purples and blues. When the more recent polls are given greater value, the impact Obama had by being the head of the ticket against McCain is erode in relation to Clinton. In other words, the McCain margins between the two are narrower in many of the states. Obama's slide then has pulled him back down to earth; to the point that there really isn't a "dime's worth of difference" between Clinton and himself.
So, now I have appeased the Clinton supporters on their methodological concerns, but probably have both camps agitated with me over how the maps now look. I have argued before that the cloudiness over who the Democratic nominee will be has suppressed some of the support the party's nominee should get given the state of the typical general election indicators (presidential approval and economy). In that regard then, this is something of a worst-case scenario for the Democrats. It does underscore how the divisiveness of the post-Texas/Ohio race has hurt them and makes that much more understandable the calls from the party's elite to decide this thing earlier rather than later.

Related: Obama's Slide Revisited

Recent Posts:
The Electoral College Maps (4/30/08)
***Please see links to past maps in the right side bar***

The State of the Race: Counting Delegates in the North Carolina Primary

The State of the Race: Counting Delegates in the Indiana Primary


Robert said...

If Obama can't stop the slide, Clinton will have an argument. Next week is critical for both Democrats.

Josh Putnam said...

Fortunately for Obama next week's contests are on "friendlier" territory than West Virginia and Kentucky will be in the next couple of weeks. Clinton would have a streak of her own if those two were up next. They are the two states, behind Arkansas, where she has the biggest McCain margins.

But yes, "friendly" as they may be, North Carolina and Indiana won't be a cake walk for Obama in this new post-post-Wright environment.

I meant to say this at the group today but your comments last week in the Moyers/Wright post were right on. Why now? The Moyers appearance wasn't bad but the Press Club stuff has killed Obama in the media.

Never a dull moment in this race.

Robert said...

Obama appears to be in free-fall. Did you see on RCP that the Insider Advantage poll has Clinton ahead by 2% in NC? The latest polls in IN are slanting her way in the 8-9% range. Losses in IN and NC by Obama would be devastating to his campaign. He is still doing reasonably well in the invisible primary with superdelegates, but that will change if Clinton does well next week. Did you see that Clyburn is telling Wright to disappear from public view?


Josh Putnam said...

Here's that link from Rob.

Josh Putnam said...

I did see that Clinton was ahead according to Insider Advantage (via Talking Points Memo) and I'm willing to give Obama the same benefit of the doubt that I gave Clinton in Pennsylvania when Obama was gaining on her there. If more polls start showing the same result, I'll put a bit more stock in that result.

Wright may be quiet, but he isn't going away. We jokingly talked yesterday of Bill Clinton and Wright getting together to have a battle over who could stick their foot in their mouth the quickest, but perhaps Wright and Willie Horton should get together and compare notes. That may not be fair, but it is becoming more and more accurate.

Robert said...

The trend lines did not look good, but they seem to be stabilizing. It is interesting that CNN is making a big deal about the Quinnipac polls in OH, PA and FL. They seem at odds with provious polls, but the interesting thing is that they show Obama beating McCain in PA while pulling close in OH and FL.

Josh Putnam said...

The cynical view is that the media pushed too hard against Obama and needs to restore some stability to the race so that it continues for as long as possible, giving them stories into the summer.

In reality, these poll numbers fit in with the news frame of the day: Andrew switches to Obama and Obama's numbers are up. It is an Obama day and the start of a new month.

Let's parse this out a bit:
January: even
February: Obama
March/April: Clinton
May/June: ????

It has gotten off to an Obama start.

We will have to see. There are some nice, new data to add into ye olde polling spreadsheet.

Robert said...

I wonder what the ditto heads will think if Clinton goes on to the nomination and then wins the election in November. Maybe they will wish that they hadn't pushed Obama over the cliff.

Yes there is an unbelievable amount of new data for your pet project today.