Landslides as re-alignments
The easiest alignment theory to dispense with is the idea that we will know it has happened by its size. The gist of the theory as proposed by Bowers is that a 400+ electoral landslide would be a re-aligning election.
Of course, this would be noteworthy, given that the Democrats have only won more than 51% of the popular vote once since FDR passed from the scene. But this is almost an historical accident as much as anything: Clinton likely would have done this in 1992 but for Perot (who actually pulled slightly more from Clinton than Bush, according to exit polling) and certainly would have done so in 1996 when he almost won 400 electoral votes, and picked up states like Arizona that had not gone Democratic since 1948. But this victory was passing, not re-aligning.
In fact, history is replete with electoral landslides that no one seriously considers a re-alignment, and many of the elections that are thought to be re-alignments were actually quite close. Consider the chart below, which shows the percent of electoral votes won in Presidential elections where one party or the other won 75% or more of the electoral vote (which is the equivalent of the 400 electoral vote threshold set by Bowers):
As you can see, there have been several massive wins for one side or the other that no one seriously considers re-alignments: Ike’s two victories stand out in particular, though at the time one could have pointed to the GOP taking the House in 1946, making big gains in 1950, and then winning outright in 1952 as signs of a realignment that would never come to pass. The Whig’s big win in 1840 over Martin Van Buren did not portend a generation of Whig hegemony, nor did Millard Fillmore’s 1852 landslide foreshadow great things for the Democrats. There are also Nixon’s 1972 win and LBJ’s 1964 win, which were not re-alignments.
Indeed, landslide victories are often followed by the collapse of the winning party; such was the case for the Democrats in 1966, the Republicans in 1974, the Republicans in 1958, and the Republicans in the House in 1982 and in the Senate in 1986. And narrowing the threshold for a landslide to 65% of the electoral vote adds little clarity, as years like 1992 and 1996 find their way onto the list.
On the other hand, several of the supposed re-aligning elections have been quite close. Not listed on the table are the 1800, 1828, and 1860 elections (of those, only 1828 involved the victor receiving even 65% of the electoral vote). 1876, which Mayhew posits as the “critical” election, was one of the closest in history, and is absent. 1896 misses the list (though the Republican victories in the 1894 midterm still represent the largest pickup in U.S. history), while 1932 was clearly a landslide, but 1968 was extremely close.
Nor is the fact that Democrats won control of the Congress in the mid-terms of any great import. Consider that in 1890, Democrats took advantage of the inept Presidency of Benjamin Harrison to take a massive 238-86 advantage in the House of Representatives (about 75% of the seats). In 1892, Republicans bounced back somewhat, picking up 38 seats, largely as a result of reapportionment, but Grover Cleveland stormed back and won a resounding 62% of the electoral college. But this did not translate to a re-alignment; in 1894 Republicans picked up the most seats in US History (+130 seats), and in 1896 they maintained a large majority while electing William McKinley, and setting in motion a pro-Republican re-alignment (arguably).
In other words, sometimes landslide wins are considered re-alignments, and sometimes they are not. It is not consistent enough for us to read anything about this election based upon the size of Obama's (or McCain's) victory. Sometimes narrow elections are not re-alignments, and sometimes they are. When it comes to re-alignments, size doesn’t really matter.