Democrats have a state legislative problem. The statistics are familiar to those of us who happen to follow numbers in politics: Despite winning the popular vote in six of the past seven Presidential elections, or whatever the best spin is, Democrats control the legislative chambers of a mere eleven states, with just eight more split between Democrats and Republicans.
My sense is political pundits and amateurs like myself usually give one or both of the following explanations for this disparity: First, that Democrats have principally declined in the formerly (and, for decades, artificially) Democratic Solid South. Second, that the party’s state legislative problem stems from something particular to Barack Obama--either from his supposed lack of attention to politics and “party-building” or from his supposed “Obama coalition” of young urban minorities who are geographically clustered, don’t vote down-ballot, and so on.
Both of these explanations seem to have some truth to them: Democrats have of course suffered enormous losses in the South, and the two midterm elections under Barack Obama were of course devastating to the party’s state legislative standing. Even together, though, they’re not sufficient to explain everything. Democrats have declined from extremely strong state legislative positions both within and outside the South (of course with a much smaller decline, and with many more exceptions, outside the South) and in both regions the decline started well before Barack Obama— whether that comes from changing preferences, changing maps and population distributions, or both.
Analyzing state legislative trends over time is difficult. There are ninety-eight partisan state legislative chambers (Nebraska being both unicameral and non-partisan), and there might not be any single best way to collapse partisan standing in all ninety-eight into aggregate numbers. I considered a few different possibilities. I could look at the number of states each party controls (and treat Wyoming the same as California?). I could look at the number of legislators each party elects (and treat the hundreds of state representatives elected in New Hampshire the same as the eighty elected in, well, California?). I could look at the number of states controlled by each party (and how to account for the occasional odd minority coalition?).
I eventually decided to look at the estimated percentage of the population in Democratic state Senate and state House districts over time, and since this is American politics, to look at the non-South and the South separately. The result can be seen below. (Note that this is not the aggregate popular vote over time--I think the numbers on that are a little less reliable.)
For these purposes, “the South” is Oklahoma, Missouri, Kentucky, West Virginia, Virginia, and everything South of them: Texas, Arkansas, Louisiana, Tennessee, Alabama, Mississippi, Georgia, Florida, North Carolina, South Carolina, so fifteen states in all. The “non-South” excludes D.C. as well as Nebraska, which has a unicameral and non-partisan state legislature, so thirty-four states in all.
On methodology: I assumed a state’s population between two Census estimates is a linear interpolation of the two (it isn’t), and that this growth is even across the state (it isn’t) and that districts thus remain equally populated (they don’t, and don’t always even start equal), and multiplied a state’s Democratic seat percentage by its estimated population to get a very rough count of people in Democratic districts. See below for more.
Again: Democrats have of course lost huge numbers of Southern legislative seats, and this is probably the greatest single change in the state legislative map over the past several decades.
That said, the fall of the Southern Democratic state legislator has been a pretty continuous one, at least since the 1980 Reagan landslide, and with the exception of the good Democratic run from 2005 or 2006 to 2010. Legislative Democrats actually lost about the same percentage of the South under “native son” Bill Clinton as they did under Barack Obama, depending on how you look at it: The percentage of the South in Democratic state Senate districts dropped 15.9% under Clinton, from 69.9% (1992) to 54% (2000), and so far dropped 13.8% under Obama, from 47.9% (2008) to 34.1% (2015). The percentage of the South in Democratic state House districts dropped 12.9% under Clinton, from 67.1% (1992) to 54.2% (2000), and so far dropped 12% under Obama, from 45.3% (2008) to 33.3% (2015).
As I said, though, it’s widely understood that Democrats have dramatically collapsed in the South, including at the state legislative level. The trends outside the South seem to be understood less widely.
The non-Southern chart is much more stable than the Southern chart (for one thing, the non-South is much larger), and Democrats certainly haven’t declined continuously, at least not by this measure. But they have declined, mostly because of the 1994, 2010, and 2014 elections.
I don’t think it’s widely appreciated how strong Democrats were after Watergate at the state level, outside the South (perhaps because Democrats mostly lost the White House over this period, except for their narrow 1976 win). From after the initial post-Watergate elections of 1973 to 1994, Democrats always had at least 54.6% of the non-Southern population in Democratic House districts, and at least 50.9% of the non-Southern population in Democratic Senate districts, with the “typical” performance a bit better in each case.
After the 1994 Republican wave, Democrats apparently only represented a minority of the non-South in state legislatures, which hadn’t been true for decades. They didn’t manage to reach even those 1973–1994 minimums until after the 2006 Democratic wave. By this population metric, the results of the good Democratic years of 2006 and 2012 were right around what used to be the Democratic baseline under Reagan and George H.W. Bush. (The median result under Reagan and Bush was about 53.5% of the non-South population in Democratic Senate districts and about 57% of the non-South population in Democratic House districts; results in 2007 were 53.9% and 57.8% respectively, while results in 2013 were 52.8% and 55.8% respectively.)
The below set of maps compares two atypical years to make this point: 1989 and 2009. In 1988, the Democrats of course lost the Presidential race by a wide margin: Mike Dukakis only won ten states, 111 electoral votes, and 46% of the popular vote. At least in the popular vote, it was about as big a loss as the Republicans suffered in the 2008 Presidential election, with an economic crash, an unpopular war, and an unpopular Republican in the White House.
The maps below currently compare to . Two-party percentages are used to color the maps; absolute percentages are used elsewhere. Click any state to get specific numbers, and click again to remove them.
Just by looking at the state legislative maps, though--even outside of the South--would you guess that 1988 was a large Democratic loss at the Presidential level, and 2009 was a large Democratic win? Democrats apparently did better in 2009 in aggregate population terms, with 60.5% of the non-South in Democratic House districts (compared to 58% in 1989) and 54.9% of the non-South in Democratic Senate districts (compared to 54.1% in 1989). Still, this improvement is pretty unimpressive considering that 2009 followed the best pair of Democratic elections in recent memory. Comparing 1989 and 2009, Democrats actually declined in at least one chamber in all but thirteen states (California, Colorado, Connecticut, Delaware, Hawaii, Illinois, Massachusetts, Minnesota, New Hampshire, New Mexico, New York, Rhode Island, Vermont).
Here are some specific states, outside the South, where Democrats have weakened at the state legislative level:
Why have Democrats generally weakened in state legislatures? That would require a more detailed analysis, state by state, which I hope to attempt in future articles.
One possibility is that Democratic strength from 1975–1994 was itself an anomaly, that the post-Watergate Democratic landslide was in some sense artificial, lasting for twenty years only because of incumbency advantages and the sheer size of the victory. That is an awfully long anomaly, though.
A related possibility is that Democrats at the state level benefitted from Republican strength at the national level, in the same way that Republicans at the state level gained during the Clinton and Obama presidencies--I believe political scientists call this the “thermostat model”> or some such. Still, Democrats were below their Reagan-era levels for most of the George W. Bush Presidency.
Democrats are generally understood to have improved in urban areas while declining in rural areas, which is generally seen as unhelpful at the state legislative level. Several states outside the South have areas that are politically and culturally similar to the South, or at least had strong local Democrats, often in rural areas, who eventually fell to national trends--southern Indiana, western Pennsylvania, southeast Ohio, rural Nevada and Arizona — while Democrats didn’t really have areas like that to lose in New England or New Jersey, where state legislative trends have been more favorable to them.
Of course, how population shifts effect political results depends, at least in part, on who’s drawing the map, and the constraints they’re under. My sense is that interpretations of the Voting Rights Act have changed over time, especially around the 1992 elections, and gerrymandering techniques are generally seen as growing more sophisticated and aggressive.
Again, I don’t see any reason why the same factors would apply to every state, and close looks at particular states, regions, and elections are probably necessary to explain everything. I’m hoping to attempt that for a few states in subsequent articles.
My main source was the state legislative historical numbers compiled by Carl Klarner and the National Conference Of State Legislatures. Historical and current state populations are from Wikipedia. The state GeoJSON and some of that code is from this block by Chris Ingrahm.
My method for the first chart was to take the decennial Census counts by state from Wikipedia, use linear interpolation to estimate the population for individual years, and to multiply that by the percentage of total seats that were Democratic (as opposed to Republican, vacant, or “other”) from the Klarner/NCSL data. As I said above, this assumes that population growth is linear between Censuses, that population growth is even across a state, and that districts are drawn to be equally populous and are thus always equally populous. All of these assumptions are wrong, but I think it probably suffices to show broad trends. Again, vacancies and independents or third-party legislators are not accounted for. Think of multi-member districts as being fractionally Democratic or something.