It's been six weeks since Virginia held its elections for the state offices of governor, lieutenant governor, attorney general, and the state House of Delegates. With the count certified by the state board of elections and the recount confirming the attorney general outcome, Democrats swept all three statewide executive offices for the first time since 1989 and won control of all five statewide offices with the inclusion of the two U.S. Senate seats for the first time since 1968. I've created 20 maps at the county/independent city level in Virginia that look at: the winning margins for all three 2013 races, comparisons to 2012 presidential performance, the distribution of actual votes rather than just percentages, the gubernatorial and three race average swing from 2009, the changing winning coalitions since 2001, and the turnout differentials between 2013 and 2012.
Click all maps for 6.5 times larger image
The above is Governor-elect Terry McAuliffe's win over Attorney General Ken Cuccinelli. Statewide McAuliffe won by just over a 2.5 percent plurality, with 6.5 percent going to Libertarian Party candidate Robert Sarvis. Using just the two-party vote makes the comparison to other races clearer and that expands McAuliffe's margin to 2.7 percent. He won the typical counties one might expect with him doing very well in the cities of Hampton Roads, the Richmond area, majority minority parts of Southside like Greensville County, Northern Virginia, and towns dominated by colleges and universities like Charlottesville. Cuccinelli did quite well in rural, heavily white counties, particularly in Southwest Virginia and the Shenandoah Valley, as well as the whiter suburbs/exurbs of Richmond and Hampton Roads. In general McAuliffe won the bigger cities with commanding vote shares such as more than 80 percent in Richmond, while Cuccinelli didn't match that in any county, let alone a heavily populated one like Chesterfield just to the south which he only narrowly won.
Some simple linear regressions looking at the correlation by county and city between the 2013 performance and 2012 presidential performance found that all three statewide races were very highly correlated with each other and with the presidential margins with R² values around .95 or greater. Additionally while this diary does not map the state legislative races the correlation was almost as high between the 2012 presidential and 2013 House of Delegates performance with an R² of .85. A simple extrapolation of the House of Delegates performance compared to 2012 president by district for the 43 of 100 where both major parties were on the ballot suggests that the Republican Party would have won the House of Delegates popular vote by roughly 5 to 6 percent if all 100 districts had both major parties on the ballot.
In summary this map collection yields a few key points:
- Democratic votes were heavily concentrated both geographically and proportionally in cities while Republican votes were spread out more in rural and suburban areas.
- Regions where each party has an advantage on average are also generally trending toward their respective party. Western Virginia is trending strongly Republican while much of rural Virginia was also trending modestly Republican. Northern Virginia, greater Richmond, and Hampton Roads are all trending Democratic in part thanks to high population growth overall and among minorities.
- For both parties votes are increasingly coming from just three regions: Northern Virginia, greater Richmond, and Hampton Roads. Democrats won two-thirds of their votes there and Republicans won half their total.
- The drop in turnout between presidential and odd-year elections hurts Democrats disproportionately more statewide than Republicans with a general 3 percent drop in total vote share statewide for Obama between the 2012 and 2013 electorates. For Democrats minority-heavy cities and rural areas see the worst drop-off, while whiter rural areas and whiter parts of cities see relatively less. The same was not true for the Republicans whose vote totals dropped by relatively similar proportions across the state. However, that correlation of turnout changes with race and population density is far from perfect.
- Winning coalitions for different races and candidates are still very similar to one another with swing being much more uniform and ticket splitting much lower than in past election cycles, which is similar to what is happening with federal elections nationwide.
Let's look at the rest of the maps below the fold.
Even though McAuliffe's win was slightly narrower than President Obama's in 2012, what makes it quite impressive is that exit polling showed that the 2013 electorate actually voted for Mitt Romney by 2 percent even though Obama won by 4 percent a year earlier. Since exit polling was not precise down to the county level I've mapped the difference between McAuliffe's vote share and Obama's 2012 vote share with the governor-elect doing better than the president in the bluer counties and worse in the redder counties. Statewide McAuliffe received just 0.6 percent less of the two-party vote than Obama.
In many counties the gubernatorial vote shares were within a percent or two of the presidential vote shares, but the regions that stand out for McAullife the college-towns such as Charlottesville, home to the University of Virginia, and cities like Arlington where he ran considerably better than Obama did. On the other hand, he did much worse than the president in south-central Virginia, particularly Danville where as shown further below turnout was uniquely terrible for Democrats. One very interesting split is how McAuliffe matched or exceeded Obama's performance in the inner jurisdictions of Northern Virginia such as Fairfax County, but ran considerably behind in further-out Prince William County. Atypical from recent elections was that McAuliffe did worse in a handful of lightly populated counties in the southwest where Virginia Democrats have run moderately ahead of Obama in recent years and where the president saw a big swing against him in both 2008 and 2012, which suggests that the region's pro-Republican presidential trend is finding its way down-ballot
Republicans did Democrats a massive favor when they chose to nominate their statewide candidates at a convention of party activists rather than through a traditional primary and that resulted in the nomination of little known and extremely controversial conservative black minister E.W. Jackson, who would prove to be unelectable (a Google search for "EW Jackson controversy" yields interesting results). Democrat Ralph Northam, a state senator from the Chesapeake Bay area, handily defeated Jackson by 10.6 percent which is the second largest margin for a Democrat statewide in the past quarter-century. Northam carried several more counties than McAuliffe and led the ticket in the vast majority of them statewide, particularly those contained in his state senate district in Virginia's Eastern Shore. Northam also won major suburban counties like Chesterfield south of Richmond and the independent cities of Chesapeake and Virginia beach in southern Hampton Roads, even though those are typically the most populous counties/cities to vote Republican. Jackson still did quite well in the heavily white, rural Republican base in Western Virginia and carried typically conservative cities like Lynchburg.
Unlike the gubernatorial map the contrast between Ralph Northam and Obama's performances is almost entirely blue since Northam won 3.3 percent more of the vote statewide. In particular he ran significantly ahead of the president in the greater Richmond area, the Peninsula region, the Eastern Shore, and somewhat surprisingly even some parts of the Shenandoah Valley and Southwest Virginia, but not in the westernmost counties that have historically been the most Democratic until recently. Jackson still managed to outperform Romney in places with large minority populations like Prince William County in Northern Virginia and Danville in Southside where turnout falls precipitously between presidential and off-year elections.
Democratic state Senator Mark Herring held a razor-thin 165 vote lead for attorney general over state Senator Mark Obenshain out of over 2,200,000 cast for a margin of 0.007 percent at the time of certification. This race underwent a recount this week that expanded Herring's lead to 907 votes, though these changes are not reflected in this diary. Ironically enough just as recently as 2005 the race for Attorney General was ultra-narrow with the Republican winning by less than 400 votes and the recount in that election changed the margin by just a few dozen votes.
In this map there are a lot more Republican-won counties and darker shades of red signifying larger Republican margins in places like vote-heavy Chesterfield County. Obenshain managed to win the Republican-leaning swing cities of Chesapeake and Virginia beach by narrow margins, as well as the nearly-black-majority City of Danville on the North Carolina border thanks to horrendous Democratic turnout. Herring still manged to win key areas of the state like the outer Northern Virginia counties of Prince William and Loudoun thanks in part to that being his home region, while Obenshain is from the central Shenandoah Valley. The attorney general results map gives a fairly good idea of what a generic winning coalition currently looks like for both parties since it is practically tied.
Herring received 2 percent less of the vote than Obama did and that under-performance was somewhat even across the state with a few notable exceptions. Herring did considerably worse in Southside and where it turns into Western Virginia and the Shenandoah Valley. Part of this is likely due to turnout (with Danville being the worst), but in places like Allegheny County one explanation might be Obenshain's native son status though he surprisingly didn't exceed Romney by all that much in his actual state senate district further north. As one might expect, Herring held up decently over Obama in Fairfax and Loudoun counties where his State Senate district is located, but his running several percent ahead of Obama in Fairfax was a rather pleasant surprise.
One fairly valuable way to measure the general performance of a party is to create an average of all the races in a particular year. This gives one an idea of what the generic level of Democratic or Republican support was with less of a distortion from a candidate' home region. Overall the statewide average vote share was 52.2 percent Democratic. Notably most jurisdictions by land-area are red, but several heavily populated ones like Fairfax County are blue and the Democratic vote is much more heavily concentrated in populated cities like Richmond and in Hampton Roads than the Republican vote is in less densely populated Western Virginia. Under the average the key swing regions of the state are the suburbs: Loudoun and Prince William counties in outer Northern Virginia, Henrico and Chesterfield counties around Richmond, and the cities of Chesapeake and Virginia Beach in Hampton Roads.
Subtracting the average Democratic vote share from Obama's 2012 vote share results in a minor advantage of 0.2 percent for the 2013 ticket statewide, but it varies considerably by region. The three Democrats ran a good deal ahead of Obama in the greater Richmond area and considerably worse in Southside where low minority turnout hurts them disproportionately. Impressively they roughly matched presidential performance in much of Hampton Roads even though it has a large minority population as well and a large pool of voters who favored George W. Bush in much higher numbers than they did John McCain or Mitt Romney. Unexpectedly the 2013 ticket did better than the president in areas dominated by colleges or universities like Charlottesville (home to UVA), Lexington (Washington & Lee), and Williamsburg (William & Mary). An interesting division was the split between the outer suburban/exurban counties of Northern Virginia with the Democratic ticket matching or exceeding Obama in Loudon County yet running considerably behind in Prince William even though the entire region has seen tremendous growth and a Democratic trend at all levels in recent years. Another significant difference between Democratic levels of support between 2012 and 2013 occurred in Arlington and Alexandria where a combination of low Republican turnout and their ticket's extreme social conservatism appears to have hurt them considerably.
(Note the scale is not linear) Another way to look at the county-level election results is not by vote share or percentage margins, but by actual vote totals. The most insightful map out of the whole collection displays the net partisan vote outcome of the three races averaged together by each locality and creates a clear illustration of just how inconsequential or important every location was for each party. Fairfax County had by far the biggest impact, providing roughly a 73,000 net votes for the Democrats which by itself is the difference between a win and a loss for Terry McAuliffe and Mark Herring. Republicans gained the most votes from very dark red suburban/exurban counties such as Hanover north of Richmond and the rural/small town counties of Western Virginia such as Bedford, Rockingham, and Augusta. Other than in Fairfax Democrats netted huge margins in Alexandra/Arlington, Henrico County/Richmond and the four inner Hampton Roads cities of Hampton, Newport News, Norfolk, and Portsmouth.
By party the vote totals reveal some major differences between the two coalitions, but also some similarities due to population distribution. Democrats won by far the most votes in the populous regions they carried of Northern Virginia, Richmond and its two adjacent counties, and the cities in Hampton Roads. Heavily Democratic Fairfax County was by far the largest single source of votes, but they also won a significant vote total even in counties they lost such as Hanover and Chesterfield near Richmond, Roanoke in the Southwest, and Spottsylvania and Stafford south of Prince William. On the other hand they won relatively few votes in heavily Democratic counties in Southside such as Brunswick and Greensville. Overall about two-thirds of the total Democratic vote came from Northern Virginia, greater Richmond, and Hampton Roads.
Republicans saw less of their vote come from the inner cities and more from the whiter suburbs in major metropolitan areas like Richmond and Hampton Roads, but unlike the Democrats a much larger portion of their total came from smaller towns and rural counties. While Fairfax County is also the largest source of votes for Republicans, cities such as Alexandria, Richmond, and Norfolk provide similar totals or less than rural counties such as Bedford, Franklin, and Pittslyvania in the west. About half of the total Republican vote came from the state's three major metropolitan areas of Northern Virginia, greater Richmond, and Hampton Roads.
The next several maps look at the changes between different state-level statewide election years with this one showing the change in two-party gubernatorial vote share between 2013 and 2009 when Republican Attorney General Bob McDonnell beat state Senator Creigh Deeds in a blowout that foreshadowed the 2010 wave. Terry McAuliffe won 10 percent more of the vote than Deeds and improved in almost all of the state with a few exceptions in Western Virginia. The darkest patch of red along the West Virginia border is because those counties were part of 2009 nominee Creigh Deeds's base and his home county of Bath saw a gigantic 27 percent swing in vote share toward Cuccinelli. The rest of the southwestern counties along the West Virginia border such as Buchanan were once a Democratic stronghold but have turned sharply Republican in recent years in large part thanks to the politics of coal and social conservatism. On the other hand the eastern, central, and northern parts of Virginia saw massive swings towards McAuliffe with him making the greatest improvement in Northern Virginia, Hampton Roads, the greater Richmond area, and the more heavily-minority parts of Southside, while he also improved considerably in many of the cities of the Shenandoah Valley but not their surrounding counties.
The swing between the three race averages of 2009 and 2013 is quite similar to the previous gubernatorial map with a statewide swing of 9.8 percent. Deeds appears to have had major coattails in 2009 in his home region with all three Democrats doing considerably worse in 2013 than the 2009 candidates in Bath and Allegheny counties, but on average they improved in all of the rest of the Shenandoah Valley. Other than that there is not much difference between the gubernatorial swing and the three race average swing with the Democrats improving substantially in Virginia outside of the west.
Looking at how the difference between the average Democratic vote share in each county and the statewide average changes between various election cycles gives produces an indication of the direction each area is trending or how it is becoming more or less likely to vote for a particular party in different years given an equivalent level of statewide support. First the trend between 2009 and 2013 which essentially shows what the average swing map did, but shifts everything several shades redder. Here the contrast of blue in Northern Virginia, Hampton Roads, and the greater Richmond with the red of Western Virginia yields a clearer view the directions each area is moving. However, since the statewide swing between those two years was so great the comparison isn't exactly apples to apples.
Current Democratic U.S. Senator Tim Kaine was lieutenant governor in 2005 when he was elected governor while the Democrats narrowly lost the other two races that year, producing a statewide average of 50.8 percent Democratic which is fairly similar to 2013's 52.2 percent. The trends in the previous map are more evident with the emerging Democratic coalition again being Northern Virginia, Hampton Roads, greater Richmond, and the more minority-heavy parts of Southside. Note that while Creigh Deeds was still on the ticket as the Democratic attorney general nominee, he didn't have such an out-sized impact on the average in Western Virginia as in 2009 and the region has still very clearly trended strongly Republican since 2005. More noticeable here is the Republican trend in those heavily white, rural parts of central Virginia and the Peninsula.
The final trend map compares 2013 to 2001 when current Democratic U.S. Senator Democrat Mark Warner was elected as governor and then Richmond-mayor Tim Kaine as lieutenant governor on his coattails. The statewide average was drastically different at 47.9 percent Democratic, but this map shows the clearest regional divergences of all. The 2013 ticket did significantly better in Northern Virginia, while running several points stronger relative to statewide in Hampton Roads, Richmond and its suburbs, and most of the college towns. To make up for it they did massively worse in Southwest Virginia and many of those counties in the darkest red were actually carried by Warner in 2001, but went for the 2013 Republican slate by over 40 percent. Unlike the previous few maps even the minority-heavy parts of Southside trended Republican since 2001 and practically all of central Virginia and the Peninsula region trended that way as well. One important thing to note though is that the population growth in the state over the last 12 years has been disproportionately in those heavily populated blue-trending counties such as Prince William while many of those dark red counties in Western Virginia are sparsely populated and also outright lost population over that same time period.
The last set of maps detail how 2013 relates to 2012 presidential turnout. The gubernatorial election had the most votes cast of the three statewide races and the totals for all candidates combined was about 58 percent of all the combined votes cast for president in 2012. The above maps that drop-off at the county level with the 133 counties/cities ranging from just under 46 percent to nearly 70 percent of presidential turnout. Counties in the lightest color saw the lowest 2013 turnout as proportion of presidential, while those in the darkest saw the highest. It was relatively lowest in heavily white southwestern Virginia as well as the minority-heavy parts of Southside and Northern Virginia, such as Manassas Park which was the lowest of all at just 45 percent of presidential turnout. Topping the list were heavily white rural counties scattered throughout the state, including some whiter suburban/exurban parts of the greater Richmond area as well as intriguingly some college-town localities like Lexington in Western Virginia. In general turnout fell by slightly more than it did statewide in typically heavily Democratic areas like Northern Virginia and the inner cities of Hampton Roads, in particular Newport News, while a relatively higher share of presidential voters showed up to the polls in heavily-white rural areas.
(Note the scale is slightly different than the previous map). Looking at the average 2013 proportion of 2012 presidential vote totals by party finds that both won a fairly similar share of their 2012 totals in 2013 at about 58 percent for Democrats and 57 percent for Republicans, but that the regional breakdown is quite different. For Democrats there was much more of a drop-off in minority-heavy parts of Southside with Danville being the absolute lowest at just over 40 percent of the presidential vote, while other cities with large minority populations such as Roanoke, Petersburg, Norfolk, and high-growth Prince William County also round out the bottom of the list. This year Democrats retained the highest proportion of the presidential vote in the greater Richmond area and surprisingly in some of the college-towns like Charlottesville though not all of them such as Blacksburg, home to Virginia Tech. The number one thing Democrats should take away from this map is the need to improve GOTV in places with significant non-white populations like heavily black Danville and Prince William County where the Hispanic population is booming.
For Republicans the change in turnout was much more even by region than it was for the Democrats with the most notable areas of relatively lowest voter participation being in the lightly-populated, swiftly Republican-trending southwestern corner of the state and more importantly in major urban areas like Arlington. By population though, most of their votes came from regions where 2013 totals as a percentage of presidential totals were quite similar to the statewide proportion, which is not surprising since their coalition does not rely nearly as much on the young, poor, and minorities, all of whom drop-off at higher rates in non-presidential elections.
The final map contrasts the previous two to see where the drop in turnout affected each party relatively more than the other. Counties in blue saw Democrats retain more of Obama's vote total in 2013 than Republicans did of Romney's, while those in red saw the opposite. Some familiar patterns emerge: turnout disparities favor the Democrats in the more established parts of Northern Virginia aside from Prince William County, the greater Richmond area, Hampton Roads aside from Hampton and Newport News, and the various college-towns like Charlottesville. Republicans retained disproportionately more as one might expect in the Southside region and some cities with big minority populations like Roanoke, but also in north-central Virginia and the Shenandoah Valley despite those areas being very white. That Democrats won a much higher proportion of their 2012 vote than Republicans did of theirs in Arlington and Alexandria, and Fairfax County suggests that the 2013 GOP ticket's staunch social conservatism turned off a lot of socially-moderate, fiscally-conservative, Republican-leaning voters in those cities.
Ultimately though it's impossible to determine turnout by party conclusively since Virginia voters do not register by party, so these past few maps can't take into account the impact of ticket splitting and swing voters on off-year vote retention, therefore comparing 2013 to 2012 vote totals remains an imperfect, yet still insightful method.
