Wednesday, September 17, 2008

How Big Was McCain's Bounce?

Hey, what's with the past tense? Well, a consensus seems to have emerged that McCain peaked -- at least in the context of the convention bounce -- some time last week. First the national polls began drifting back toward Obama and now we have begun to see hints of a similar pattern on the state level. But how and where did McCain bounce following the GOP convention in St. Paul? As has been our custom here at FHQ, we keep tabs on these shifts with periodic examinations of the changes to our weighted averages. Typically that has meant a monthly wrap up post, but the two conventions straddled the line between August and September, making that a difficult enterprise. The trends on the state level during the first two-thirds of August didn't really seem to favor either candidate despite talk of Obama's continued slide in the polls. Much of that could be attributed to the fact that many of the mountain west states were not polled during that time. In other words, we did not have a complete, state-by-state understanding of what was going on. Those gaps have been filled in the post-convention period.

Using the electoral map FHQ posted on the eve of the Democratic convention as our baseline, we can compare where the race was on August 24 to where it is now (...or was as of last Saturday*). I should note that we are talking about changes to our average and by its nature that means that the shifts are, on the whole, kind of small. To get a large shift, one of two things has to happen: 1) a number of polls indicates a distinct trend toward one candidate in a state or 2) one outlier poll completely changes the average in state that has been polled very sporadically. North Carolina is the best example of the former. The latter is well illustrated by a trio of states: Louisiana and West Virginia moving toward Obama while Vermont moved toward McCain. And the movement in these three states is due to a decrease in a substantial lead. Vermont's average for example shifted nearly 6.5 points based entirely upon a new poll (one of three total from the state) that had Obama ahead just 19 points instead of a number in the mid- to high twenties. Keep this in mind as you look at the movement across each state.
[Click Map to Enlarge]

And what we see is an awful lot of McCain yellow. The Arizona senator put some distance between himself and Barack Obama in a lot of traditionally Republican states, from the mountain west and into the South. In fairness though, McCain also made up ground in more competitive states like Ohio, Minnesota, Michigan, Wisconsin, Nevada and Oregon. He also increased his small advantage in Virginia. But Obama made strides in several states as well. Most notably, he gained ground in Indiana and Missouri and extended advantages in Pennsylvania, Iowa, and New Hampshire.

Pennsylvania being a gain state for Obama during the McCain bounce period may raise some eyebrows. Hasn't there been a decided down trend for the Illinois senator in the polls conducted in the Keystone state? Yes, but over the summer when Obama was enjoying a noticeable lengthening of his lead in Pennsylvania polling, our average nearly inched out of the toss up distinction, but never did. While other electoral college analyses moved Pennsylvania into safer categories for Obama, it never budged from the toss up distinction here. [Awfully braggy, aren't we, FHQ?] That has a lot to do with the fact that there was already a massive amount of polling done in Pennsylvania during primary season. That kept the average for the Keystone state rooted in the 3-5 point range for Obama for the most part. So, the average removes most of the volatility from poll to poll, keeping Pennsylvania anchored to a certain equilibrium, and in the process mutes some of the movement that otherwise looked to be heading in McCain's direction.

Let me close by focusing on the swing states. States like Alaska, Montana, North Carolina and North Dakota all moved off the board -- perhaps not permanently, but they are off for now -- during this post-convention period. Among the remaining eleven toss up states, though, McCain gained in seven (Nevada, New Mexico, Colorado, Michigan, Ohio, Virginia and Florida) while Obama inched up in four (Missouri, Indiana**, Pennsylvania and New Hampshire). I have asked the question before, but was the bounce in the national polls distributed evenly across all states? No. McCain shored up his base in some traditionally Republican states, but also seems to have edged out Obama among the battleground states as well. What does that mean, though? If this is, in fact, McCain's peak, then pulling to within eight electoral votes (273-265) is close but not quite good enough. However, if external factors intervene and one candidate has the momentum down the stretch, those toss ups may break in concert in one direction on November 4.

*The map charts the changes to FHQ's weighted averages in the states over that period (August 25 - September 12). Last Friday was the back end boundary because that was the cut off between a week of intense polling and a weekend where the releases dwindled. In the time since the week end we've seen a subtle shift back toward Obama. So to capture the full McCain bounce, it was incumbent upon FHQ to include only those polls that were conducted and released during the Arizona senator's peak period in the polls.

**Of course, Indiana's one poll during this period was released between the Democratic and Republican conventions. That poll favored McCain, but the margin was smaller in that Howey-Gage poll than the previous two polls conducted in the Hoosier state.

Recent Posts:
The First Presidential Election Votes Get Cast Tomorrow

The Links (9/17/08): Debate Edition

The Electoral College Map (9/17/08)


Robert said...

Looks like the bounce is over, at least nationally. I think it may be a little early to see how things are settling out in the states. The financial situation appears to be taking its toll on McCain-Palin. I expect that to start showing up in the Rust-Belt states next week that tilted to McCain with the Palin effect.

Scott, your assessment of sticking with the running game by Obama rather than the razzle-dazzle of McCain appears to be paying off for the time being. In a close race, though, it will be the external factors that could affect the ultimate outcome.

Josh Putnam said...

Here's that RCP link from Rob.

It is too early to see if the regression to the mean that we've seen in the national polls this week has extended to the states. But I'll assess that more fully next week. I'll look at the changes we will have seen from last weekend to next Friday. That may not give us the clearest picture, but it'll be the best we'll able to do before debate season kicks off and potentially disrupts the equilibrium again.