15 June 2004

American Partisanship

The Times ran an interesting piece this weekend on the nature of American partisanship.  This is pretty typical of the gist of the piece:

Do Americans really despise the beliefs of half of their fellow
citizens? Have Americans really changed so much since the day when a
candidate with Ronald Reagan’s soothing message could carry 49 of 50

To some scholars, the answer is no. They say that our basic
differences have actually been shrinking over the past two decades, and
that the polarized nation is largely a myth created by people inside
the Beltway talking to each another or, more precisely, shouting at
each other.

These academics say it’s not the voters but the political elite of
both parties who have become more narrow-minded and polarized. As Norma
Desmond might put it: We’re still big. It’s the parties that got

Just because a state votes red or blue in a presidential election
doesn’t mean that its voters are fixed permanently on one side of a
political divide or culture gap. The six bluest states in 2000, the
ones where George
W. Bush fared worst – Rhode Island, Massachusetts, New York, Hawaii,
Connecticut and Maryland – all have Republican governors. Even
California went red last year when Arnold Schwarzenegger, a moderate
Republican, became governor.

Most voters are still centrists willing to consider a candidate from
either party, but they rarely get the chance: It’s become difficult for
a centrist to be nominated for president or to Congress or the state
legislature, said Morris P. Fiorina, a political scientist at Stanford
and senior fellow at the Hoover Institution.

Now the article sums up the conventional explanation for the increasing partisanship of the electorate as follows:

Why, if the public is tolerant, would the political elites be so angry?
One reason given by Professor Fiorina is the decline of party bosses,
who promoted centrist candidates because their patronage systems
depended on winning elections, and the corresponding rise of
special-interest groups, who are more concerned with candidates’

I spoke with political scientist friends at Berkeley who actually study this
stuff (like Mo Fiorina), but they disagree with his conclusions,
partially out of a technical methodological question (which has huge
implications, actually), and partially out of a very different
understanding of the chain of cause and effect.

First, the method problem.  The focus of this article is on the
state and national levels, but people elect Congress from their
particular districts.  So we’re dealing with two different levels
of aggregation, as it were.  On the larger level, people are more
moderate, but on the smaller level — the Congressional district —
voters are actually quite intensely partisan.  Think of it this
way.  Pretend the political landscape actually is a
landscape.  So it looks like a topographical map, where you can
see the elevation of any particular point, but from which you can also
calculate an average (mean) elevation above sea level.  So imagine
you’ve got a landscape, and it has peaks and valleys. 

there are peaks or valleys, their depth or height indicates the
degree of partisanship.  But the average level for a picture like
the one above show a fairly middle of the road elevation.  But if
you lay a grid horizontally over the picture, the squares will likely
contain more peak or valley than the opposite, and overall you get a
sector that is primarily peak or valley.

This somewhat gives you an idea of the methodological problem. 
When we talk about the country as a whole or even a state or region as
a whole, we smooth out the partisan bumps that exist.  If you laid
a partisan map over the country, similar in approach to the graph
above, you’d find wide and radical differences between parts of the
country.  Mississippi would be a fairly high peak, while the San
Francisco Bay Area or Boston would be a very low valley.*  So the
map of the U.S. would be pretty bumpy.  But if you look at the
country as a whole and apply the statistics and measurements that
describe the country or even a region on average and apply that
backwards to the very small level of congressional districts, you get a
much smoother, more moderate topography.

*I choose the peak and valley for the party based on how we normally
treat party idenitfication numerically, with Republicans as +1 and
Democrats as a -1.  No implication about relative worth exists,
because my own biases would probably reverse putting Republicans on the
sunny peaks.

Second, in the U.S., most congressional districts
are highly peaked or “valleyed”, indicating a high degree of
partisanship in any particular district.  Most districts are
comfortably Republican or Democrat, and so are a high peak or low
valley.  This, according to my political scientist friends, occurs
for two reasons.  First, legislatures and parties in them have
drawn comfortably uncompetitive districts, where one party dominates by
eight or ten percentage points over the other.  Second, within the
parties, voters just do tend to be fairly partisan.  For some
reason (my friends are working on the answers to this, as we don’t
quite know yet), voters act in intensely partisan ways, especially when
they live in comfortably partisan districts.  It’s only when a
district has a competitive balance between the parties that we see
“moderate” candidates emerge, and we have an extraordinarily low number
of competitive congressional districts in the country today.

Interestingly, I learned that primary voters, who are often reputed to
more partisan than their party-fellows at large don’t differ
significantly from those party-fellows.  In other words, it is
something of a myth that we get more partisan party candidates because
people who vote in primaries are more extreme than those who vote in
general elections.  Republicans and Democrats who vote in
primaries are not significantly more ideological than Reps or Dems at

So the NY Times article provides a very interesting viewpoint on the
partisan wars of the country, but perhaps a little too
Pollyanna-ish.  The data that matters, from individual
congressional districts, indicates that we are an intensely divided
people.  We may even out as a country, but we don’t vote for our
politicians (excepting the President) by country, and so it’s not a
reasonable way to look at our current political atmosphere.

ADDENDUM: David Brooks’ column today
posits another reason for the partisan divide in today’s politics —
it’s based upon a conflict between professional and managerial
classes.  As often, it’s an interesting read….

Be Sociable, Share!

Comments are closed.