Snowman on Election Day

To follow up on Karl Rove and the phantom realignment, in general I think everyone needs to be much more cautious in their theorizing about political trends and the deep causal origins of electoral outcomes. It's fairly easy to construct a little narrative about postwar electoral politics that fits the data in a plausible way. So easy, in fact, that you should begin to get suspicious -- several different narratives seem to fit the bill.

What should really trouble you, however, is that if you look at presidential elections from 1948-2004, Democrats have won the popular vote seven times and Republicans have won it eight times. This means that on the list of plausible narratives about post-war electoral politics is that the outcomes are completely random and that DRRDDRRDRRRDDDR is just a sequence like you might get from flipping a coin. Which isn't to commit myself to the view that the outcomes really are random (who knows?), but merely that one ought to be very cautious about embracing some Grand Narrative or another.

Comments

The outcomes in terms of the popular vote may be random, but I'm not so sure (I guess I could test this ... you can do runs tests on the internets) about the outcomes of the electoral college.

The thing is, as we know, elections are not won by the popular vote ... certainly not by a majority of people voting for you (as I pointed out way down in the "Halfway" thread). What you need is a plurality of people in a weighted majority of states. And the GOP has been able to get out the vote in enough states (as well as to piss off enough people to get them not to vote for either candidate), to win elections.

We should be cautious about embracing narratives that don't hold up statistically, but something does seem to be happening. But we can't blame/credit Rove for the Southern Strategy, can we?

Posted by DAS | August 13, 2007 10:54 AM

It's just possible that policies, not elections, are the things people should be looking at for trends.

Posted by SomeCallMeTim | August 13, 2007 11:00 AM

The comment to which I referred above:

here.

It's in reference to this comment.

Note that if the most centrist candidate (rather than the candidate "perceived" ... Brad does include that very key qualifier ... to be most at the center) were to always win (the electoral vote), Kerry would have won in 2004. Only a person of the mindset "all Democrats are nutso liberals" or "anybody who isn't gung-ho about the war must be a dirty hippy" would view Kerry as less centrist than Bush ... especially by 2004.

Unfortunately, the so-called liberal media thinks this way ... with the results I describe in my comment ...

Posted by DAS | August 13, 2007 11:03 AM

The Atlantic piece on Rove mentions the "Grand Narrative" that he put forth amongst the Washington elite to account for his GOP successes--it centered on how he (Rove) was the only one who correctly understood the significance of William McKinley in 1896. Really.

We certainly should be cautious about embracing narratives that don't hold up statistically, but I suspect that the people who want to believe in the kind of utterly fatuous crap typified by Rove's story will find a way to do so no matter what.

Posted by James Gary | August 13, 2007 11:07 AM

I'd say the mostly even split b/w Dem. and Repub. wins just shows that both parties are more committed to what political parties are supposed to be good at (winning) than to any particular principles or policies (which are supposedly more important).

After the Dems figured out the American people didn't trust them on national security, they have worked hard to out-hawk the hawks. When the Repubs realized that refusing to offer any form of prescription drug Medicare benefit would open a huge election issue for the Dems, they passed a fax, expensive entitlement program.

On the whole, I think this is good (not specific outcomes, but the trend): when America changes, the policital parties change with it so they can remain competitive.

Posted by JWR | August 13, 2007 11:25 AM

Except if it was random you would see longer strings of the same party.

In 50 flips of a coin you are likely to see heads or tails 7-8 times in a row. One party has never held the presidency that long.

Definitely not random.

Posted by Gully | August 13, 2007 11:32 AM

Sorry, you would expect runs of 5-6 in a row...

Its still early here in CA.

Posted by Gully | August 13, 2007 11:36 AM

Gully: no, it could still be random with those statistics. Just not binomially random with p=q=0.5. There's lots of kinds of "random".

Posted by Wugga | August 13, 2007 11:36 AM

Yes. Party choice in Presidentia elections is pretty much a coin flip. Because the Parties are strategic in their choice of candidates for executive positions, and the candidates for those positions are strategic in their self-presentation.

Mitt Romney has been governor of Massachusetts! The governor of California is a "popular" Republican!

Political realignment is about establishing people's default choice of Parties in legislative elections, and, not incidentally, the country's ideological center of gravity.

The truth about the strategic positioning of the Parties vis a vis one another is that each is trying for a 60/40 split, nationwide. And, since 1994, the two Parties have been getting 49-49, with the advantage going to the Republicans due to sheer ruthlessness.

A really good narrative is going to make something out of the country's drift toward social atomism, and the decline in ethnic and racial prejudices as well as social and civic associations, the rise of television, the decline of labor unions, the rise of the plutocracy and the explosion of CEO compensation, Media consolidation and the creation of the Republican Wurlitzer and a lot of other topics.

The bottom line, which the Atlantic article on Rove skirts, is that Bush/Cheney/Rove have proven criminally bad at governance. They put all the authoritarians into one Party, and as Krugman points out recently, although leadership of authoritarians is easy, it does require lying and stupidity at a level, which is repulsive to everyone else. Democrats are gaining, in reaction.

But, the Democrats have really poor institutional support. The unions are very weak. The Media is dominated by numnutz pundits hired by corporate Republicans. Racial and ethnic political activism repulses as many people as it attracts.

Posted by Bruce Wilder | August 13, 2007 11:37 AM

Wugga,

But he compares it to flipping a coin...

You would definitely get more than 3 in a row. But I expect you know more stats then I do.

Posted by Gully | August 13, 2007 11:38 AM

Wugga does have a point. I think we've been taking too much of a shorthand in our language. I certainly had binomially random with p=q=0.5 in my mind when I wrote my post, but I guess what I said could apply to even testing the hypothesis of whether the events were independent as well ...

Posted by DAS | August 13, 2007 11:38 AM

I think that Rove accomplishments were enormous and self-limiting. I would compare him to Attilla who in rather short time created vast barbarian empire that after his death dissolved into small pieces.

The methods applied by Huns were good for cobbling together a vast coalition, but the latter was cemented by a vast stream of war spoils that could not be secured forever.

The analogy is apt, because one of the Rove's method was escalation of spoil system to the degree perhaps unknown in 100 years. Another was "energizing the base" which is a non-financial form of spoil system. And, as in the case of Huns, the idea of attending to mundane government tasks were alien to people like Rove and DeLay.

Posted by piotr | August 13, 2007 11:45 AM

Of course, realignment arguments aren't primarily about the presidency. If you look at congressional elections from 1948 to 2004, Dems dominated Congress pretty much continuously until 1994, except for a brief period in the early 1950s and the loss of the Senate in 1980, which swept in a group of weak Republicans easily bumped when they came up for re-election six years later. Since 1994, Republicans have controlled Congress [albeit not by much] most of the time, and even now are only a Lieberman switch away from controlling the Senate; 2006 notwithstanding, it remains unclear that the realignment that took place in 1992-1994 [it really did]is coming unstuck. Of course, since it produced a rough partisan parity, marginal shifts in the present system can make a big difference in elections. The political system growing out of the Civil War/Panic of 1873 didn't actually produce Republican dominance, but an even balance characterized by intense partisanship and two of the three presidential elections in American history in which the Electoral College went against the voters; sound familiar? The sort of stable party-systems that realignment theorists study don't necessarily entail single-party dominance, nor are they inconsistent with parties swapping the White House; what matters is whether the partisan allegiances of voters remain stable across multiple elections. There's plenty of evidence that it does; certainly your own interest in the notion that under-30s are establishing strong, stable Democratic loyalties lets realignment theory in the back door.

Posted by David | August 13, 2007 11:51 AM

Presidential elections are not random at all. The party in power almost always wins re-election unless there is either:

1) a bad economy
2) a bad war
3) a bad scandal

So Bush the Elder loses in '92 because of recession. Clinton wins in '96 because of the strong economy and lack of other negative factors. Bush the Dumber "wins" in '00 in reaction to Lewinsky but only barely illustrating that the scandal was not that serious. Kerry almost beats Bush in '04 due to Iraq despite the superficially good economy. If the Presidential election had been held in '06 instead by which point only the Republican base continued to believe Iraq is a success, Kerry would have won pretty easily.

Posted by Ron | August 13, 2007 11:58 AM

I'm not buying that looking at Presidential popular vote outcomes since WWII is going to tell you very much about trends (or the lack of them) in American electoral politics. "DRRDDRRDRRRDDDR" doesn't account for the fact that Democrats dominated Congress for generations following the New Deal, and leaves out significant midterm elections like 1994 and 2006. And, in fact, it doesn't really tell you that much about the Presidential races, either. I appreciate the notion of cutting out all the noise and looking at results only, but it's just too superficial a metric to understand what's going on.

Posted by Jake H. | August 13, 2007 11:58 AM

Argh, David beat me to it.

And Ron has a good point as well--Presidential incumbents (including, usually, VPs) win unless there are sufficient negative factors which create an opportunity for the challenger.

Which is all to say, you can't really believe that Presidential elections on the whole are random unless you believe that individual elections have random results. Does anyone really believe this? You could argue for a lack of long-term realignment-style trends, but that's not randomness.

Posted by Jake H. | August 13, 2007 12:06 PM

Gully, the Dems won 5 in a row from 1932-1948. And then the GOP won 5 of 6 from 1968-1988. Does this help to confirm a null hypothesis of randomness?

Posted by nbt | August 13, 2007 12:12 PM

I think all you really need to know about Karl Rove and realignment is summed up in the quote in TNR 8/22/02:

"I look at this time as 1896, the time where we saw the rise of William McKinley and his vice president, Teddy Roosevelt," Rove declared. "That was the last time we had a shift in political paradigm."

The only problem is, Roosevelt didn't run with McKinley in the 1896 election - he didn't become Vice President until after 1900.

Posted by Ethel-to-Tilly | August 13, 2007 12:22 PM

First, the Republicans won the popular vote in 1960 so the sequence is "DRRRDRRDRRRDDDR".

Second, as others have said, realignment isn't about presidential elections. Legislative majorities are more stable and reflect deeper public sentiment about policy. Presidential elections are just momentary popularity contests.

Third, for DAS: If you really think Kerry was the more centrist candidate, you're missing out on what happened in 2004. Kerry was the most leftist candidate for president we've ever had in this country. Kerry wanted gay rights and legal (medical) marijuana and fewer wars and major environmental laws with teeth and universal health care and gun bans and higher taxes. Kerry made McGovern look moderate.

Bush is a moderate Republican. Bush supported new welfare programs and more federal involvement with education and bigger federal spending.

Confusing spinelessness on the greatest issue of the day from Kerry with moderation and confusing corruption from Bush with conservatism tells more about the writer than about the candidates.

That Kerry almost won in spite of his radicalism and a decent economy reflects deep and widespread public dissatisfaction with Bush in 2004.

Posted by Adult | August 13, 2007 1:09 PM

Kerry was the most leftist candidate for president we've ever had in this country.

This comment betrays a deep lack of knowledge of both politics and history.

Posted by dob | August 13, 2007 1:39 PM

Adult,

I guess it all also depends on how you define what is the center. I'd say compare Kerry's stated viewpoints on the issues with most people: you'd find that most people, lying in between us moonbats who didn't wanna go into Iraq and who support gay marriage and those who think we should nuke Iraq and force gays back into the closet, are pretty much where Kerry was, and not where Bush was.

Nu? If liberalism is higher taxes, then one could argue Ike was more liberal than Kerry. If liberalism is gay rights, well ... then Goldwater, at least at the end of his life, sure was liberal (how did he feel about medical marijuana).

And environmental laws and anti-war stances make one a liberal? My great-grandparents were, in their day, considered right-wingers. They didn't much care for activist gummint. And that included when the gummint said we should go to war; my great-grandparents hated commies, but as far as they were concerned, 'Nam was just a plot to increase the sales of bullets (remember what party affiliation Ike "beware the military-industrial complex" Eisenhower had ... that used to be something about which conservatives fretted). But the one place where they liked gummint? Environmental regulation ...

My point is that how you define the political spectrum is itself an issue. If you take the long historical view in which all but the hardest reactionaries support viewpoints on certain social issues that were considered moonbattery not so long ago, then yes, perhaps Kerry was the most liberal candidate evah.

But coming from that historical point of view -- not to sound like a whig which I really am not -- there are certain historical trends which would place Kerry as a centrist who neither wanted to over-accelerate those trends nor to reverse them. From that point of view, Kerry was no more liberal than Ike. And GW Bush was a hard-core reactionary.

However, in that long historical view, as mentioned above, certain of Kerry's positions actually would have been considered more conservative than Bush's positions (and now that Bush & CO "conservatism" is failing, the many conservatives who are jumping ship are wont to point out how Bush & CO are hardly conservative and love to classify Bush's ideology as a form of evil liberalism).

OTOH, ignoring the historical point of view and looking at how things play out now, under what metric was Kerry further away from the political center than was Bush? And is that center all that liberal? Indeed, in as much as what is labeled "centrist" often benefits a nascient neo-aristocracy, it can hardly be called liberal at all, can it?

Posted by DAS | August 13, 2007 1:54 PM

Just for those who wish to compare it to coin flipping:

With 15 flips, there is a ~20% chance that the longest streak would be 3 in a row. It is certainly not freakish.

Extend it back 5 more elections, and the most likely biggest streak of coin flips is 5, which is exactly what happened.

There have been 37 elections since the founding of the Republican party. With 37 coin flips, the most likely longest streak is 6. The Republicans won 6 straight elections from 1860 to 1880, though there were a lot of complications.

I would assume that there are dynamics involved that make electoral outcomes less random than coin flips, you can not glean anything like that from the results. From the birth of the republican party, you can replace the Ds and Rs with heads and tails and the results do not even raise an eyebrow statistically.

Posted by Njorl | August 13, 2007 1:56 PM

First, the Republicans won the popular vote in 1960 so the sequence is

Er, it can be argued that Nixon really won the popular vote in 1960, because votes for the Democrats in Alabama were, in fact, votes for a slate of electors which included both Kennedy and unpledged electors. But normal counts of the 1960 election give Kennedy about a 100,000 vote plurality

Posted by John | August 13, 2007 2:13 PM

OOps, math off slightly above...not much though.

With 15 flips, 4 (29%)straight is most common, but 3 straight isn't far behind(22%)

With 20 flips 4 straight is most common(26%), but 5 straight is close(25%)

With 33 flips (33 elections with 2 major parties and no civil wars or occupations) 5 straight is most common (26%) coinciding with 5 straight dem victories.

With 37 flips (since inception of Republican party) 5 straight is most common, but 6 straight is very close, both around 26% likelyhood. There were 6 straight Rebublican victories spanning the Civil war.

With 51 flips assuming the first 2 were non-partisan, and the rest were Democratic or Democratic-Republican party vs "other", 6 straight is still the most common longest streak at 26%.

It is all very consistent with coin flipping. Granted, you could get 51 straight heads while flipping a coin, but somewhere around 20 or so you'd suspect you had a 2 headed coin.

Posted by Njorl | August 13, 2007 2:17 PM

Congressional elections are much more reflective of broad trends because the number of quasi-independent races is large enough to support some degree of aggregation. And the 06 election delivered a larger majority in the House to the Dems than the 94 election gave the Reps, so it's pretty safe to say the House ain't switching back next fall. The Senate favors the Dems because of the larger number of R seats defended plus the general political climate. But the Presidential race is subject to so much psychological manipulation and political tactics that calling it a "popularity contest" isn't as inaccurate as I'd like. The odds of a Dem victory are not much better than 50/50.

And the "Republicans" lately aren't so much conservative as they are corporatist and authoritarian. So much so that the word "Republican" seems an ironic label for them. But liberal they're not, despite the best efforts of "Republican"/authoritarian loyalists to shift the opprobrium onto the Dems.

Posted by STS | August 13, 2007 2:48 PM

STS,

I'm not as sanguine as you. I suspect the Dems. in 2006 were put their by the electorate for specific reasons. And they've been screwing up. So I'm not sure how long the Dem. majority in Congress will last.

Posted by DAS | August 13, 2007 3:02 PM

Er, it can be argued that Nixon really won the popular vote in 1960, because votes for the Democrats in Alabama were, in fact, votes for a slate of electors which included both Kennedy and unpledged electors. But normal counts of the 1960 election give Kennedy about a 100,000 vote plurality

It is a fact that Nixon won more popular votes outside Alabama than Kennedy. It is a fact that Nixon electors won more popular votes within Alabama than Kennedy electors. Objective observers can draw their own conclusions from that.

The "normal" counts you have seen published that give Kennedy a majority count all votes for Harry Bird electors as votes for JFK. That is the only way to put Kennedy in the lead. No plausible form of averaging will give the result that Kennedy led the popular vote.

Kennedy's popular vote victory is akin to the WMDs we found in Iraq and Saddam's plan behind 9/11. It's just a nicer story than telling the truth.

Posted by Adult | August 13, 2007 3:08 PM

DAS,

I agree the Dems have been disappointingly weak in pushing against Bush on Iraq and civil liberties, but what is your scenario for a Republican tidal wave in the House? Every newly elected Democrat in 06 is beaten by a hastily rebranded "Republican" challenger?

I think the (karmic) punishment for Dem congressional weakness would be more likely to take the form of a "Republican" presidential victory.

Posted by STS` | August 13, 2007 3:18 PM

"DRRDDRRDRRRDDDR"

Say it out loud and I think you'll see the real issue here.

Posted by George Tenet Fangirl | August 13, 2007 3:48 PM

People are right, the sequence of victories looks periodic, not random. People tend to get tired of one party when that party is in power for a while, and they "throw the bums out".

By the way, this:
"DRRDDRRDRRRDDDR"

Say it out loud and I think you'll see the real issue here.

is possibly the funniest thing I have ever read.

Posted by Mr. Noah | August 13, 2007 4:27 PM

Adult,

Your contention that Nixon got more popular votes outside of Alabama than Kennedy is not true. Kennedy's popular vote total outside of Alabama was 33,902,681, while Nixon's was 33,870,176. (The popular vote total for Harry Byrd and other 3rd party candidates was 495,147. This includes Harry Bryd's 116,000 plurality in Mississippi.)

Moreover, even if you factor in that the Democratic electors in Alabama included unpledged electors, it is still true that the in Alabama, the Democrats got more popular votes for the presidential race than the Republicans did, even if those Democratic votes were not specifically pledged to Kennedy.

Posted by eltoro | August 13, 2007 7:15 PM

What should really trouble you, however, is that if you look at presidential elections from 1948-2004, Democrats have won the popular vote seven times and Republicans have won it eight times. This means that on the list of plausible narratives about post-war electoral politics is that the outcomes are completely random and that DRRDDRRDRRRDDDR is just a sequence like you might get from flipping a coin.

That makes no sense whatsoever. Realignment theory has always centered on the finite nature of realignments, usually lasting a generation before it falls apart. The fact that there has been a nearly-even split in the partisan outcome of Presidential elections over the past sixty years indicates that the political system has had several seismic shifts between the dominance of one party over the other, not that there has been no shift (much less that past elections were nothing more than a coin flip).

In any event, any doubt that partisan realignments really do occur would do well to compare the national map of states that went for Carter in 1976 (or JFK in 1960) with states that went for Kerry in 2004.

Posted by Steve Smith | August 14, 2007 4:11 PM