| Nickolas also produced a table of the underlying figures, which I've resorted according to popular vote margins:
If we take this table, slice it up, and match the numbers with the electoral college maps, then whole story becomes a bit more vivid. It also throws further light on my earlier discussions of electoral maps over the last century, here and here.
Election Margins -0.5% to 2.4%
The first slice contains two slices within it - first from -0.5% to 0.7%, second from 2.1% to 2.4%. The reason for putting these two slices together will become obvious from the associated maps:
 |
There are obviously two distinct slices here. 1960, 1968 and 2000 are truly close, close elections. So close that loser "won" in 2000. In contrast to these elections, margins of 2 points or more seem quite comfortable-especially after the fact. But if we look at both slices together, we find that we have all the electoral maps in two separate groups that I described in my last diary, "Electoral Map Typology". Bush's two elections comprise their own group, which is an offshoot from the main sequence of Republican-dominated maps. The main sequence moves from Republican landslides with a sprinkling of Democratic states (1972, 1980, 1984, 1988) to Republican dominance over a Southern-centered Democratic Party, with the Democratic base growing stronger until it finally becomes a majority. Bush's two victories form an offshoot that's unique, since it centers on the South, previously the Democrats' base.
The other three maps here are from another anomalous group-at least it was anomalous at one time. There really was no precedent for Kennedy's 1960 victory map, but Humphrey in 1968 and Carter in 1976 both came up with geographically quite similar maps, and Clinton latter consolidated the patchwork into his own distinct map type, winning a very similar pattern of states in both his elections. So what we have here, in these two slices of close and super-close elections, are distinctive configurations that first appeared in 1960 and 2000. The first recurred again in 1968 and 1976, the second recurred in 2004. That's three of five elections from 1960 to 1976 (interspersed with two landslides, one Democratic, one Republican) and two of two elections from 2000 to 2004.
These are what truly close elections look like, and it certainly is conceivable that John McCain could win such an election this year, with a map very similar to 2000 and 20004. But right now, that is not what the polls are showing. They are showing something closer to an Obama victory that generally falls into the next category.
Election Margins 4.5% to 8.5%
Here, again we combine two slices to give a little bigger picture. Here we see one pattern that crosses over the divide-Clinton's two geographically similar vicotries. But we also see significant differences in crossing the divide.
The first slice, Truman's 1948 election and Clinton's 1992 election, is the one that Nickolas cites as most directly comparable. The dominance of tighter polls from Gallup and Rasmussen may make his ballpark of a 5-point Obama lead seem a bit high, but with Obama putting states like Virginia, North Carolina, Indiana, Missouri, North Dakota and Montana into play, there does seem to be good reason to see the maps for these two quite different elections as broadly indicative of what we might see come November. These are very comfortable Electoral College victories-landslides, if you will. But they are not geographical landslides, landslides that isolate the states supporting the losing candidates into a few small islands of support.
Crossing over into the second slive, we start to see signs of geographical, as well as Electoral College landslides. In 1988, Democrat states were limited to four small enclaves-one of them a single small state. In 1944, the Great Plains were the only geographical region that the Republicans held. In 1996, the Republicans carried a significant number of states, but they were divided from each other into three blocks-and again, one of those was just a single small state. The divide between Truman's 1948 victory and Roosevelt's 1944 victory is where I earlier drew the line over what I was willing to call a landslide. That divide shows up here as well, and it seems well justified in terms of geography, and the electoral vote margin. However, there is no doubt that Truman's victory was a very solid one.
In contrast, from here on out, we're only talking further degrees of landslide, and the elections fall fairly neatly into three clumped groups.
Election Margins 9.7% to 10.9%
Empirically, this is the minimal level for an unambiguous landslide.
The three elections in this slice are widely separate in time and flavor. Roosevelt's 1940 margin was a decline from even more overwhelming margins in the two previous elections. Eisenhower and Reagan were both elected in repudiation of their predecessors' difficulties, particularly with foreign affairs.
Election Margins 15.4% to 18.2%
This is where crushing landslides begin.
Two of these elections were re-elections, and two were first term. Interestingly, the first term landslides were back-to-back-Hoover in 1928, FDR in 1932.
Election Margins 22.6% to 24.3%
These landslides only exist to make those above look puny.
Conclusion
Now, no one's saying this is going to be 1936. But it could be something close to 1932 or 1956 if McCain has a total meltdown at the debates. But even without that, something in the 4.5-8.5% range is quite doable, and would result in an Electoral College landslide of something like 200 votes. Right now, a total blowout for Obama is just as likely as a squeaker victory for McCain-if not moreso. |
