Did the Ground Game Matter in the Colorado Senate Race?

Did the Ground Game Matter in the Colorado Senate Race?

The recent Colorado race raises questions about how Bannock Street Project tactics could hold up in future contests.

Seth Masket
Jan 5, 2015

One of the most closely analyzed elections from 2014 has been Colorado’s U.S. Senate race, in which first-term Democratic Senator Mark Udall narrowly lost to Republican Representative Cory Gardner. The New York Times provided a close look at this contest just last week, examining which voters turned out and which did not.

Offices can be run very differently across campaigns, and it’s not obvious that more offices means a more effective turnout effort.

Part of the reason this contest has received such scrutiny is because it was a test of the so-called Bannock Street Project, a locally focused get-out-the-vote (GOTV) effort pioneered by the Obama campaign in 2008 and refined by Michael Bennet’s Senate campaign in Colorado in 2010. (Bennet was one of very few swing-state Democratic incumbents to survive that year.) The architects of the Bannock Street Project were confident they could boost Democratic turnout again, as they had in 2010, saving Udall’s seat even though he was trailing Gardner by a few points in the polls.

Relatedly, this was also an important race because it seemed that Republican strategists had finally learned the importance of voter turnout efforts. After McCain, Romney, and Colorado Senate candidate Ken Buck had been defeated in Colorado while making few specific turnout efforts, the Gardner team devoted time, money, and office space to its GOTV campaign. How much did these efforts matter?

One helpful way to look at the Gardner-Udall┬árace is to compare it to the Bennet-Buck Senate race from four years earlier. As in 2014, this was a close match in a Republican-trending year, with a Democratic incumbent fighting to maintain a recently acquired seat. Below is a scatter-plot comparing the two contests. The horizontal axis shows the Republican’s vote share in 2010; the vertical axis shows it in 2014. Each data point is a Colorado county. The green trend line shows what the 2014 vote would have been if each county had voted just as it did in 2010.


Leave a Reply