Thursday, November 1, 2018

Some thoughts on NC Legislative Battles Shaping Up on Nov. 6

With today's release of a new Meredith College Poll, they find that both congressional and state legislative generic ballot questions have Democrats favored by 10 points over Republicans.

This is certainly an eye-popping number, and may reflect the Democratic enthusiasm level (on a scale of 1-10, probably safe to say a 12), but Democrats are facing two important barriers to their blue wave in a blue-moon election cycle in the Old North State.

First, in North Carolina and in the nation, Republicans have had historical advantages when it comes to voter turnout. In a previous post, I noted the turnout rates of registered voters, by party registration, in North Carolina's past elections, going back to the 2002 mid-terms:



With the exception of 2006's Democratic wave year, when registered Democrats tied with registered Republicans at 40% turnout each, each mid-term sees a generally higher percentage of registered Republicans showing up in mid-term elections.

The second distinct barrier for Democrats to overcome will be the strong partisanship of so many of these districts, thanks to partisan gerrymandering by the Republicans at both the U.S. House and state legislative districts. One way to envision this dynamic of partisanship within the districts is to look at the relationship between how a district voted for President Trump at the top of the 2016 ballot and what percentage of the vote a Republican legislative candidate got in that same district in that same year.

First, for the 13 U.S. House districts in North Carolina, look at the closeness of the fit of each district in terms of how Trump's vote percentage predicts the percentage of vote for the U.S. House Republican candidate:


The three Democratic districts--the 1st, 4th, and 12th--are all at the bottom, while the 10 Republican districts are at the top. For example, in the 2nd Congressional District, held by George Holding, Trump received 53 percent of the district's vote to Holding's 57 percent. In the 9th, Trump received 54 percent of the vote to then Robert Pittenger's 58 percent, and in the 13th (a completely new district with new candidates), Trump received 53 percent to eventual winner and Republican Ted Budd's 56 percent. 

Yes, there's not a perfect 100 percent relationship between the two figures: Trump tended to do slightly worse than the Republican congressional candidate, but the 'fit' of the districts (granted only 13) is a strong 0.976: meaning, Trump's vote could explain 97.6 percent of the vote for the Republican congressional candidate.

So, if we accept the close relationship between Trump's performance and Congressional Republican candidates performance in the same district, here's where it gets interesting: what happens if Trump's 'popularity' within the district were to drop 5 points? In the 2nd, a drop of 5 points would put Trump at 48 percent--where is Holding's percentage? In the 9th, a drop of 5 points would be put Trump at 49 percent, and in the 13th, a drop of 5 would also put Trump at 48 percent. And if the connection is that tight to the president by these Republicans, is there a subsequent drop in their potential numbers as well?

At the state legislative level, here's the relationship between Trump and state senate Republican candidates:



Again, the relationship is measured at a 0.953, meaning Trump's performance in state senate districts again closely align with the Republican state senate candidate's performance.

For the state house, the relationship isn't as strong as in the state senate or congressional analysis:




Still, 82 percent of the state house Republican candidate's vote can be potentially explained by Trump's performance in the state house district.

And while Republicans bought themselves an insurance policy in crafting these districts, the danger was the 100-year flood that wipes out the natural advantage and comes to claim the policy. And if these generic ballot measures show a swing against the president (38 percent presidential approval per Meredith's poll, significantly below the national average of mid-40s), tell me what the swing is against the president in North Carolina and we can start to potentially 'plot' and 'assign' various house and state senate districts into groups, by simply subtracting a number from their 2016 performances.

In the following charts, I took the 2016 performance and classified the districts as: those with Trump's performance below 45 percent; those with Trump between 45-50 percent; those with Trump between 50-55 percent; and those districts with Trump's performance above 55 percent. For most analysts, the range of 45-55 indicate a competitive district that could potentially swing one way or the other. 

Then, I did the same for the new 2018 districts (since we had to undergo redistricting yet again) and compared those to the 2016 numbers.

Finally, I took two scenarios: a 5 point swing (may a really good night for Democrats) against the president in the districts and how many districts would "move" from one group to another, and then an 8 point swing (something that would be unheard of and labeled more of a tsunami than a wave). 

Now, I will repeat this again later, but this is not a prediction of how exactly these districts would behavior. There could be many potential surprises on November 6, and some may be very district specific that leads to the outcomes--again, not a prediction but a potential 'simulation' of what might happens, all else being equal (and this year, that's not necessarily the case). 

First, the North Carolina state senate districts:




You'll note that for the senate districts, under the 2016 map, 4 would be considered 'underwater' for Trump's performance and probably relied on local candidates getting the requisite bump up needed for victory. In the 2018 districts, 5 are under the 50 percent threshold for the president's performance.

But going into the two simulations of 5 points and 8 points move against the president, potential 12 senate seats become precarious with a five point swing, and 14 senate seats become precarious with an eight point swing. 

Next, the state house, using the same classification and simulation numbers:



Again, with a five point swing against the president's performance, Republicans would need to rely on their candidate's over-performance when it comes to potentially 14 House districts, while with an eight point swing, it's 28 districts. Again, this is not predictive, but rather just a simulation, with all things being equal, of what might happen if the president's performance becomes a potential drag on Republican legislative candidates, and a Democratic wave is powerful enough to crest over the partisan walls built up against these districts. 

Of course, we'll see, soon enough, how the environment shapes up for potential swings and North Carolina's blue-moon election. For some more analysis, I'd encourage folks to check out the North Carolina Free Enterprise Foundation and Jonathan Kappler's work on partisan district analysis of state legislative seats, as well as a recent blog post over at NC Politics on congressional, state senate, and state house analyses. See you at the polls and Tuesday night.