Friday, December 31, 2010

Making Census sense

Last week, the Census Bureau confirmed what many Washingtonians had been anticipating ... that we will gain a tenth Congressional District when the 113th Congress convenes in January 2013.  Throughout 2011, the Washington State Redistricting Commission will ponder, propose, and create the new boundaries for our nine ten Congressional Districts and our 49 Legislative Districts.  In a very real sense, they will redraw the political map of the state.

I’ll be watching the Commission’s deliberations with great interest.  In my work, I’ve made some use of geographic information systems (GIS) to examine patterns of trauma transport and hospitalization, so I have a very basic understanding of the tools they’ll use to construct the districts.  The choices they make will have important effects on Washingtonians for a decade.

I had hoped to complete a post describing the Census Bureau’s methodology for reapportionment and demonstrating how it is that Washington earned a new seat, but as usual I’m running behind [SPOILER ALERT:  WA-10 was the 432nd district apportioned].  Instead, I’ll present a few random observations on the national picture of apportionment over the last century or so.  The data on trends in state populations, percent change, density, and district counts since the 1910 Census were downloaded from the Census Bureau.

• Washington’s count of districts last changed in the 1990 Census; WA-09 was, in fact, the very last district on the priority list in that Census.  The state has never lost a district since 1910.  We had five districts in that Census, moving to six in 1930, seven in 1950, and eight in 1980.
• Sixteen other states share Washington’s good fortune of never forcing incumbent Representatives to face off against each other.  Six of those are special cases.  Alaska (1, since statehood in 1959), Delaware (1), Idaho (2), New Hampshire (2), and Wyoming (1) have retained the same number of districts throughout.  Hawaii was assigned a single seat until the 1960 Census took effect, and has been apportioned two seats ever since.  Eight of the rest are western states—Arizona (1 when it gained statehood in 1912 to 9 in 2010), California (11 to 53), Colorado (4 to 7), Nevada (1 to 4), New Mexico (1 to 3), Oregon (3 to 5), Texas (18 to 36), and Utah (2 to 4).  And it’s no surprise that Florida (4 to 27) makes the cut.  The surprise in this group is Maryland (6 to 8), perhaps reflecting the growth of the federal bureaucracy.
• On the other side of the apportionment distribution, over half of the states have never gained a seat in any Census since 1910.  We’ve already accounted for six of them (the steady states) in the previous bullet point, but there are 22 others whose House delegations have shrunk without ever having increased.  Many of those states are in the Midwest or Rust Belt—Illinois (27 to 18), Indiana (13 to 9), Iowa (11 to 4), Kansas (8 to 4), Minnesota (10 to 8), Missouri (16 to 8), Nebraska (6 to 3), North Dakota (3 to 1), Pennsylvania (36 to 18), South Dakota (3 to 1), and Wisconsin (11 to 8).  Though their decline hasn’t been as precipitous as some Rust Belt states, portions of the Confederacy and its borders haven’t fared well either—Alabama (10 to 7), Arkansas (7 to 4), Kentucky (11 to 6), Louisiana (8 to 6), Mississippi (8 to 4), and West Virginia (6 to 3).  New England’s share of the House has fallen consistently, with Maine (4 to 2), Massachusetts (16 to 9), Rhode Island (3 to 2), and Vermont (2 to 1) in this category; Connecticut is the only New England state that has ever added a seat since 1910, going from 5 to 6 in 1930 before reverting to 5 in 2000.  Finally, Montana lost its second district in 1990 (it’s been the most populous single-district state ever since, falling short of regaining that seat by only five spots this time).
• Washington now ranks 13th in population, jumping over Indiana and Massachusetts in the last decade.  It’s unlikely to move up the ranks in the next ten years, though, as #12 Virginia is over 1.2 million above us at the moment.  With still-growing Arizona up to 16th and only about 300,000 behind us, Washington might drop a spot in the 2020 Census.
• With 101.2 residents per square mile, Washington is the 25th most densely populated state.  It surprises me somewhat that Washington, over half of which is basically empty, is more densely populated than Iowa (54.5 per square mile, ranked 36th), Minnesota (66.6, 31st), and Missouri (87.1, 28th).  As it has been since surpassing Rhode Island in 1970, New Jersey is the most densely populated state at 1195.5 per square mile.
• While the nation’s population increased by 9.7% in the last decade, Washington grew by 14.1%.  That’s nowhere near the state’s highest between-Census growth rate.  Our population more than doubled in the decade between 1900 and 1910, from 518,103 to 1,141,990 (+120.4%).  The Evergreen State has grown by more than 20 percent in three other Censuses:  1950, 1980, and 2000.  In fact, except for the 1930 Census, Washington’s population growth rate has always exceeded the national rate.
• Looking at both the levels and the timing of state population changes reinforces some of my earlier observations about the sizes of House delegations.  For instance, only once has Arizona’s population risen by less than its 24.6% in 2010, topping out at 73.7% in 1960.  Conversely, Iowa’s population has never increased by more than the 8.1% recorded back in 1920, and it even lost population in both 1910 and 1990.  No wonder the Hawkeye State has surrendered nearly two-thirds of the House seats it held in 1910.
• In the long term—the century between 1910 and 2010—Texas and Pennsylvania have essentially switched places.  They had 18 and 36 House seats, respectively, in 1910, compared to 36 and 18, respectively, in 2010.
• Since peaking at 45 House seats after the 1930 and 1940 Censuses, New York has lost at least two districts every decade.  That’s seven consecutive Censuses!  I can hardly imagine the bloodshed in the 1982 election, when the Empire State dropped from 39 to 34 CDs.
• On the other hand, imagine the fun in California over the years, as it added 9 seats in 1930, 7 in 1950, 8 in 1960, 7 in 1990.  The just-completed Census represents the first time since 1920 in which California’s House delegation failed to increase.  California actually came close to losing a seat in 2010; its 53rd district was #434 on the priority list.  By the way, that 1920 Census outcome merits a very, very large asterisk.  For reasons that have eluded my research thus far, Congress decided not to reapportion after the 1920 Census.  One of the items on my to-do list is to simulate a reapportionment for that Census, using present-day methodology.  I also want to hit the library, or at least Google, to find out why there wasn’t a reapportionment in 1920.

Two issues dominate the discussion of Congressional redistricting in Washington:

Where will the new district be located?

and

How will the new districts cross the mountains?

Those questions are linked, of course.  It’s clear that the dry side of the state does not have enough population for two districts, so at least one of the other eight will have to occupy space (and represent Washingtonians) on both sides of the Cascades.  Given the requirement for sensible communication and travel within a district, it’s likely that either the I-90 corridor or the Columbia River counties (or both) will carry a Congressional district into eastern Washington.  We’ll know better after the Census detail tables are released.

In addition to what promises to be a fascinating look at the intersection of demographic technology and political power application as Washington constructs a 10-district Congressional map, I’ll be watching how the Redistricting Commission handles redistricting at the state and county levels.  There won’t be any changes in the number of state legislators or county council members, but the outlines of every legislator’s district will change.  We won’t really know what that means until the precinct-level Census counts are released by the Census Bureau.  If the 2000 Census is any guide, that should happen in late March.  I’ve glanced at the state’s by-LD population estimates for 2010, which are probably quite accurate.  They show significant population differences across the current districts, which by definition had very similar population bases in 2000.  There will be significant reconfiguration of the LDs, all over the state.  How, or whether, these changes will affect the state’s political climate remains to be seen.