- The Civitas Partisan Index provides a snapshot of past voter behavior in state legislative districts.
- In this November’s midterm election, CPI was 90 percent accurate in predicting state legislative election outcomes.
- Legislative “upsets” – races that defied their CPI category – shed light on the many factors that can influence state legislative races.
The Civitas Partisan Index (CPI) assigns partisan leanings to the districts of the North Carolina General Assembly. CPI uses voting data from the most recent Council of State elections – governor, attorney general, and other state-wide offices – at the precinct level to assign each district a score. Those scores are then divided into eight categories – Safe, Likely, Lean, and Swing for the two major political parties.
CPI ratings do not consider many factors that can influence elections, such as campaign financing, voter turnout, and candidate backgrounds. CPI does, however, provide useful information about past voter behavior in state elections. With that in mind, contests that were not correctly predicted by their CPI category can still provide useful insights about voter characteristics within those districts and races.
The 2018 legislative elections were held under newly-drawn districts. CPI ratings for those districts suggested that the North Carolina Senate Districts would favor Republicans with a 31-19 margin. Of the 50 state senate seats, CPI correctly predicted the outcome of 46 contests.
CPI predicted that Republicans in the Senate would retain their supermajority, but by a slim 2-seat margin. Pending a final vote count in Senate District 09, it appears as though the Democrats will have broken the supermajority by 1 seat.
In the North Carolina House, CPI categorized the districts as a 75 to 45 Republican-Democrat split. This would have maintained the Republican-held supermajority by the same 4-seat margin, although some of the specific seats on each side would have flipped control.
Democrats performed much better than was predicted by their 2016 voting outcomes in certain House districts. They broke the Republican supermajority by picking up a net of 9 seats. Interestingly, two of the three Swing Democrat seats went to Republicans, but Democrats made up the losses in other races. They even managed to flip one Likely Republican seat, held by incumbent Rep. John Bradford.
What can we learn from the seats that defied their CPI categorization? First, turnout matters. This year, North Carolina had a “Blue Moon” election, meaning it was a midterm election with no U.S. Senate race. With no statewide headlining contest, candidates and parties had to motivate their supporters to get to the polls.
This election had unusually high turnout for a midterm, with 52.4 percent of registered voters casting ballots compared to an average of 42.8 percent for the previous five midterms. Even though he wasn’t on the ballot, President Trump was undoubtedly a factor in driving out progressive voters in some key state legislative districts in urban and suburban areas. This cost Republicans several of their seats in the General Assembly. It was largely those seats that were “upsets,” meaning they were not correctly predicted by their CPI.
Some have speculated that the urban and suburban Republican losses were caused by a changing demographic in those areas – that those districts have become more progressive since 2016. This may be a factor. If so, it would be good news for Democrats if they hope to maintain and expand their legislative gains in the next election. Republicans, on the other hand, should hope that they were just unable to sufficiently motivate their supporters in those districts to turn out in this election. If that is the case, they could make up ground in 2020 simply by reaping the benefits of Donald Trump being on the ballot again. If Republicans hope to sustain voter support in the long term, though, this election should motivate them to play fewer political games and stick more closely to their conservative values.