Electoral Maps 1896-2004--Contiguous Regions

[UPDATED 6/20/07 7:05 PM To Reflect notapipe's Correction]

I was going to title this "Much Ado About Just A Little," but wanting at least one or two folks to take a look, I thought better of it.  But the title would be quite apt, I thought.  Until I finished writing it.  Turns out, there was a bit more here than I first thought, at least as a point of departure for further reflection.

I've looked at all the presidential electoral maps from 1896 to 2004, analyzed how many contiguous regions the states won by each party's candidates fall into, and come to the conclusion you'd expect, if you thought about it: the losers' states are more fragmented than the winners.

But I hope that putting so much work into it will at least yield a moment or two's reflection on the seemingly obvious to carry something of value away.  And if you'd like to see how that hope turns out, then join me over the flip.

Okay, the logic first, then the mechanics, then the results, then the reflection.

Got it?  Good. Let's go!

The Logic

I got to thinking about the whole "red state/blue state" thing, and my knowledge that the states were roughly reversed from their 1896 alignment--1896 being the big "realigning election" from just over a century ago.  And I started wondering about how the blocks of states had shifted around over time.

The logic of this was quite simple: I wanted a way to put our current situation into some kind of historical context, a way to back up a bit and look at our recent "red state/blue state" map through more objective eyes.  Without any preconceptions about what it would tell me, I just wanted a different perspective, to see if it might shake loose some ideas.

I decided to start with 1896, because (a) it was an historically significant election and (2) shortly after that all the territories went away, leaving 48 contiguous states.

The Mechanics

I started thinking in much too fragmented, needlessly sophisticated terms--treating the states like cellularautomata.

After a moment or two's reflection, I said to myself, "Hey, why make it so hard, when there's no obvious reason to?  Why not just look at contiguous regions?"--Meaning, blocks of states that border one another.

In the simplest case--only looking at the lower 48 contiguous states--all the Democratic states are in one block, each bordered by at least one other Democratic state, and all the Republican states are in another block, each bordered by at least one other Republcan state.  For simplicity sake, I thought of this as the basic, or "zero state," so instead of counting 1 contiguous block for each side, I decided to subtract one from the total number of blocks.

Thus, in 1964, when Barry Goldwater won 5 Southern states plus Arizona, that counted as 1, and LBJ won everything else--with Florida isolated by the 5 Southern states Goldwater won--that counted as 1 as well, for a total of 2.

There were just two complicating factors: the Great Lakes and the territories before 1912.  I decided to treat blocks separated by territories as "semi-contiguous" for a value of 0.5.  And I decided--somewhat arbitrariy--to treat none of the states across the Great Lakes from each other as contiguous except for Michigan (in light of the Upper Penninsula) and Minnesota.

I then calculated the total blocks, as well as the number of blocks for all candidates who won one or more state, and the difference between the winner and loser of the major national parties. (In 1912, the Bull Moose Party took second place, otherwise it was always Democrat or Republican.)

The Results

        Tot   Win   Lose  3rd  1st-2nd
1896     1     1     0          1
1900     1.5   0.5   1         -0.5
1904     0     0     0          0
1908     0.5   0     0.5       -0.5
1912     4     0     3     1   -3
1916     4     1     3         -2
1920     0     0     0          0
1924     0     0     0          0
1928     1     0     1         -1
1932     2     0     2         -2
1936     1     0     1         -1
1940     3     0     3         -3
1944     3     0     3         -3
1948     5     1     3     1   -2
1952     1     0     1         -1
1956     0     0     0          0
1960     7     4     3         -1
1964     2     1     1          0
1968     7     3     4         -1
1972     0     0     0          0
1976     5     1     4         -3
1980     3     0     3         -3
1984     0     0     0          0
1988     5     2     3         -1
1992     6     3     3          0
1996     4     2     2          0
2000     5     1     4         -1
2004     2     0     2         -2
Avg:     2.6   0.7   1.8 1.0   -1.1

The Reflection

What we see is totally unsurprising: the winning party has just over 1 less contiguous block on average than the losing party.  There was only one election were only two elections in which the winning party had more contiguous blocks than the losing party--1896 and 1960.  This is unsurprising for two reasons:

(A) The purely mathematical reason: The winning party usually will win more states.  The party winning the most states will be less likely to have them split up into smaller blocks.

(B) The cultural/political reason: American national politics have always been regionally based to a large extent.  Both parties have tended to encompass both liberal and conservative tendencies--unlike most other industrial democracies--but with fairly stable geographic core areas, where they have been politically dominant for decades at a time--or, for the Democratic Party in the South, for a period of over 150 years.

Because of this, it makes sense that parties rise to power by expanding their appeal beyond their regional cores, and fragmenting their opposition into disparate blocks who have less in common that holds them together, but are primarily united in what they're against.

Of course, our recent history is anomalous in at least two ways: First, the rise of mass electronic media, from radio networks through broadcast tv, then cable, and now the internet, has reduced (though hardly elminated) the capacity to specifically regionalize campaigns.  Second, the parties are more ideologically homogeneous than they have been in a very long time.  Whether these factors will outweigh commonalities with the past is anyone's guess.

But the contiguous states lens helps to remind us that we are not divided into coastal elites vs. the heartland: this account totally ignores the Democrat's significant Great Lakes presence.  And this sort of division is not all that uncommon in our history.  Nor is it uncommon for maps to change significantly between periods of relative stability.  The relatively unified winning blocks that Bush II had contrast sharply with the fragmented blocks assmbled not just by Clinton in 1992 and 1996, but by Bush I in 1988 as well.  And this, in turn, followed 4 straight elections in which the winning blocks totalled just one excess contiguous block.

Indeed, the winners' number of contiguous blocks is a very crude measure, varying only between 0 and 1 in every single election between 1896 and 1956--a period in which control of the presidency changed party hands four times.  Compared to that long run, our recent maps hardly seem set in stone.  Indeed (forgetting Florida entirely) if Gore had eeked out a win in New Hampsire in 2000 (as Kerry did 4 years later), Gore would have won with 3 excess contiguous blocks, while Bush lost with none--something entirely unprecedented.

My point is simple: This somewhat unorthodox, but hardly exotic way of looking at electoral maps both confirms conventional wisdom at one level:  the dominance of a geographically broad unified base is a strong indicator of electoral success, and undercuts conventional wisdom at another: the supposed stability of the GOP's red state dominance is purely illusory.

What emerges as more stable, however, is the basic fact of contiguous state block voting.  The exact configuration of such blocks may vary greatly over time, but the blocks themselves persist. Winning isolated states is far less significant than expanding the size of existing blocks, which in turn can depend on developing regional thematic appeals.

A Parting Thought

I conclude with a final reflection that is purely speculative at this point: If the GOP has a lock on most of the South (which an Edwards candidacy could disturb somewhat) this is arguably not the case elsewhere.  As one indication, it's quite conceivable that if Kerry had responded quickly and vigorously to the Swift Boat Liars attack, he could have won in 2004, not by winning Southern states, but by escaping the "effete New England liberal" label. The question for 2008 is how can a Democrat capitalize on the negative residue of the Bush years, not by simply running negatively against it, but by creating a contrasting positive.

One possible answer echoes a tact that Kerry himself should have taken to disarm the largely specious attacks on him as a flip-flopper: emphasizing flexibility as both a sign of strength and source of stability: like a tree with strong roots, outward flexibility is a key to longevity and perseverence.  Properly articularted, such a theme has great potential in the Great Plains, and the Mountain West as well as the Mississippi River-bordering states that Clinton won in 1996, but Kerry lost in 2004.  All these states have physical rural roots that resonate with such a message, as well as long-term state-level cultural and economic history that resonates as well.  This stands in sharp contrast with the relative inflexibility and intolerance of the GOP's Southern base.

This is, of course, a purely speculative suggestion at this point.  But something of this sort is precisely what Democrats ought to be thinking about now: developing cultural images and associations that have regional resonance differing from that which the GOP's Southern base demands--and that provide a foundation for the sorts of policy innovations that actually produce results.

Tags: 1896-2004, Electoral Maps, Presidential elections (all tags)



Re: Electoral Maps 1896-2004--Contiguous Regions

If you had some pictures, this post would be even better.

by DaveB 2007-06-20 03:24PM | 0 recs
Re: Electoral Maps 1896-2004--Contiguous Regions

Does this then come down to the Midwest swing states like Ohio and Missouri, maybe Wisconsin and Minnesota?  The Northeast states seem like they will break for the Dems, Pennsylvania is a toss up I think, but that's a solid blue block.  Southern states are still solidly Repub, but maybe we can make inroads in Western states like Montana and Arizona.  Interesting diary, I tend to like  or agree with your comments.

by Kingstongirl 2007-06-20 05:52PM | 0 recs
Generally, Yes

Though with a bit more breathing room, unless we really screw up.  I think it's possible the 2008 map may look somewhat more like 1996, especially if Edwards is the candidate.

This is something that I don't see discussed that much (but, then I tend to tune out a lot of partisan diaries): for all the talk about how Obama is reaching out to Republicans, it's hard to see how he wins a lot more states that way.  I see him maybe doing a lot better in the states he does win, and probably picking up enough to win. But I don't see him making deep inroads, the way Clinton did, and the way that Edwards possibly could--though probably not as much.

by Paul Rosenberg 2007-06-20 06:20PM | 0 recs
Re: Electoral Maps 1896-2004--Contiguous Regions

I think you got 1960 wrong. Backwards, in fact. Kennedy had five contigous units (since Mississippi didn't go for Kennedy, but for Byrd and Thurmond), while Nixon had four.  Here's a map, with Kennedy in red.

And if you count DC, In 72, McGovern had two units.  (Ditto for Mondale in 84.)

by notapipe 2007-06-20 05:52PM | 0 recs
Good Catch!

Now corrected.

I went through it several times catching mistakes I'd made. But the point comes where you look directly at your mistake, and say, "Yup, 2+2=5!"

As for DC, I should have mentioned leaving it out, I guess.  It just really does seem geographically anomalous.

by Paul Rosenberg 2007-06-20 06:11PM | 0 recs
Re: Faithless electors, third parties, etc.

There have been three adjustments to the two party map over the course of American history.  Recently the most common exception has been the faithless elector.  This has had two periods of "popularity":  the longest has been from 1956 to the present.  In 7 of the 13 elections at least one elector defected (2004, 2000, 1988, 1976, 1972,1960,1956).  In all 7 of these elections, an elector from the losing side bolted.  In 1960, one Nixon elector defected and 14 others voted for Harry Byrd, 8 as an unpledged slate and 6 others elected on an individual basis as "Democratic" electors from Alabama).  In 1864 one Republican elector did not vote.  The other period when electors voted for "other" candidates was from 1808 through 1820 with faithless electors in 1808 (6 of them) and 1820 (1)and 2 "blank votes" cast in 1816.

The second adjustment, by far the largest, has been the performance of third parties.  In 1824 there was only one party.  Democratic Republicans included Jackson, John Quincy Adams, Crawford and (amazingly) Clay.  Jackson got the most popular votes and the most electoral votes but second place John Quincy Adams had the significant support of fourth place Clay (one state one vote if it goes to the House) in the House of Representatives and won.  In 1832, it was the Whigs who were divided (and lost).  Real third parties got electoral votes in 1856, 1860, 1892, 1912, 1924, 1948,and 1968.  These were regional parties with the big exception of Theodore Roosevelt's Progressive Party of 1912.  TR reeled in 88 electoral votes compared to the Republicans 8.  Many of the third place parties were southern (1860, 1948, 1968).  The populists were national but drew their electoral votes from the Mountain West and Plains for the most part.

No Democrat has served as President without winning a large number of electoral votes from the south.  Only Cinton served without a majority of southern states and he won six of thirteen in both 1992 and 1996.  Of course, many of us believe that Al Gore was elected President and carried just one southern state (Florida).

Most surprising to me was the 19th century fascination with experimenting with the electoral college formula on a state by state basis.  Some states electors were named by the state legislature (SC did that through 1860).  Others had a state wide election and indidvidual elections on a congressional district-by-congressional district basis.  In 1892 at least one state had a fusion ticket consisting of three Democrats and one Populist that was elected.  Another state tried the novel method of electing by congressional district with the two state wide electors chosen by the congressional district winners.  The potential splits in Maine and Nebraska really are nothing new under the sun.

The most dangerous moments in the history of the electoral college are kind of obvious.  In the first election, George Washington was the unanimous choice for President but John Adams received just 34 of 69 votes for Vice President.  Shortly thereafter, the runner-up came from the losing party (Jefferson, 1796).  Of course in 1800, Jefferson's runningmate Burr received the same number of electoral votes as Jefferson and the rival party got to choose the President in the House.  It took 36 ballots (and the eventual support of Hamilton) for Jefferson to get a majority of the states.  In 1860, Lincoln was elected over two regional rivals and a weak national opponent despite have no support (and I mean no support) in the south.  In 1868, loser Horace Greeley died after the election but before the electoral college met.  If that had happened to a winning candidate what would have happened?  We almost knew when in 1932 Chicago Mayor Tony Cermak stepped in to take a bullet meant for FDR in Miami.  That was IIRC in December. Ironically, Cermak had supported Al Smith for the nomination in 1932.      

by David Kowalski 2007-06-20 06:24PM | 0 recs
Re: Cermak and Roosevelt

Cermak was shot in February 1933.  At least some people think that Cermak rather than FDR was the target.  

by David Kowalski 2007-06-20 06:41PM | 0 recs
Re: Cermak and Roosevelt

But IIRC in those days the President was not inaugurated until March 4 or so, so FDR was still President-elect when John Nance Garner almost moved up from VP-elect -- or whatever would have happened, a puzzlement even now as you suggest.

by Woody 2007-06-21 05:30AM | 0 recs
Mechanics question

Minor point/question

"And I decided--somewhat arbitrarily--to treat none of the states across the Great Lakes from each other as contiguous except for Michigan (in light of the Upper Penninsula) and Minnesota"

Not sure I completely get this.  You mean Michigan and Pennsylvania are not contiguous because they are across Lake Erie from each other, right?  That makes sense.  But Michigan and Minnesota do not share a border, the UP shares with Wisconsin.  Did you still count Michigan and Minnesota as contiguous, because of the UP?  Being from Michigan I guess I would generally agree that the Yoopers are culturally similar to Minnesotans, moreso than Trolls (those from the LP, not GOP people reading this site), but Yoopers are a small portion of the Mich population.

But I'm not sure Mich-Minn as contiguous makes any more sense than Mich-Ill as not contiguous.  In fact, I'd say it makes less sense, but I wouldn't count either pair as contiguous.  

Just saying...probably only effects one score someplace.  Like adding another unit to Kennedy in 1960.

by The lurking ecologist 2007-06-20 09:48PM | 0 recs
Like I Said, It Was Arbitrary

(And no doubt has its roots in my childhood obsession with maps, which began at a very early age, where all sorts of forbidden desires to remake the world are born.)

Michigan shares a water boundary with both Minnesota and Illinois, true.  But in addition, Michigan includes a small island, Isle Royale, just off Minnesota's northeastern tip.  Altogether, I just felt that there was a genuine cultural contiguity between the two, but not between Michigan and Illinois.  (Gary, Indiana most emphatically sees to that!) I only think it affected a couple of elections.

by Paul Rosenberg 2007-06-20 11:22PM | 0 recs
Re: Like I Said, It Was Arbitrary

So why didn't you major in Geography at UCLA?

by adamterando 2007-06-21 05:54AM | 0 recs
Re: Like I Said, It Was Arbitrary

Old news by now, but Michigan's trolls are much more populous than yoopers and more culturally similar to Illinois than Minnesota.  But that's OK.  It's your analysis.

Isle Royale has no permanent residents except wolves and moose and a scientist and ranger or two that may stick around in the frigid winter to study them.

by The lurking ecologist 2007-06-25 12:08AM | 0 recs


Advertise Blogads