On Tuesday night, voters in New Mexico’s House District #46, located in Santa Fe county, removed three term Representative Carl Trujillo and replaced him with a 31 year old political newcomer Andrea Romero. The race featured #MeToo issues on one side and accusations of misusing public funds on the other. The final result was close (52.7% Romero — 47.3% Trujillo). Below the surface though, the results of this race have a lot to tell us about who is voting in 2018 and what it could mean for the midterm election.

House District #46 has an interesting history. It was represented by former House Speaker Ben Lujan, father of US Congressman Ben Ray Lujan, for more than three decades. While the Speaker was driving to Albuquerque every day in 2010 for cancer treatments, Carl Trujillo ran a negative campaign that fell just short. I did polling in this race. In a late tracker, the race narrowed but I had Lujan ahead outside the margin of error. After a weekend of negative stories, the race closed further and the Speaker won by two points. It remains my greatest professional regret that I did not convey how dire the situation could become and instead focused my memo on the positives in order to keep the campaign focused on what they needed to execute to win.

Two years later, after redistricting, Carl Trujillo ran against the then Santa Fe Mayor David Coss. The Mayor said he planned to continue his job as Mayor if elected. That raced turned decidedly negative, and after some late spending on behalf of Trujillo by Jay McClesky using Republican Governor Susana Martinez’s campaign PACs (see mailer here), Trujillo won 52.1% to 47.9%. While, on the surface, this seems close, it masks a large urban/rural divide.

In the rural towns in the north, Trujillo won by a large margin. In the city of Santa Fe (the second map above), and the towns just north of the city, Coss won by significant amounts. The only close area was the precinct where Chupadero is located. While Trujillo was unopposed in 2014 and 2016, the more progressive candidates did worse in the northern/Trujillo precincts than in the southern/Coss precincts. In the 2014 primary for Governor, progressive Alan Webber beat Gary King by a wide margin (56.8% to 21.2%) in the Coss precincts. Webber lost 39.2% to 22.6% in the Trujillo precincts. Bernie beat Hillary 54.5% to 44.5% in the Coss precincts. Hillary won 52.1% to 47.9% in the Trujillo precincts.

Despite all the drama in the race, the basic structure of the results were similar to 2012 albeit with a big exception. Trujillo got 70.6% in the precincts he won in 2012 and got 70.2% in this same precincts in 2018. Andrea Romero did 5.6 points better than David Coss in the Coss precincts.

A greater share of voters cast a ballot in the Trujillo precincts in both 2012 and 2018. The turnout advantage was significant in 2012 as the race would have been tied if the same share of voters participated in each region of the district. Since 2012, however, the number of registered Democrats fell by 85 in the Trujillo precincts and increased by 487 in the Coss precincts. Combined with an increase in turnout rate from 32.7% to 39.7%, 799 more people voted in 2018 in the Coss precincts. This represents a 29.5% increase in the number of voters. In the Trujillo precincts, turnout increased, but the voter base shrunk, producing just a 10.5% increase in the number of voters. This change in regional composition cost Trujillo the race. If Coss increased his support to Andrea Romero’s 67.7% level in his precincts, he still would have lost given where the voters were distributed and who decided to participate in 2012.

Now what does this tell us about voters in 2018? In the early vote in this district, 16.1% were new primary voters. They voted in neither the 2014 gubernatorial primary nor the 2016 presidential one. Another 24.3% voted in just 2016 and not 2014. This was similar to what happened statewide where just 54.1% voted in 2014 and 16.9% were completely new. The share of new voters is significant but figuring out who they are in advance is practically impossible as the turnout rate among these voters was exceedingly low (4.2% statewide, 6.2% in district #46).

The electorate is different from the electorate when Barack Obama was president. New people are voting and the distribution of them is not consistent either geographically or ideologically. In all of Santa Fe County (see chart below), the higher the increase in the number of people voting from 2014 to 2018, the better Bernie Sanders did in 2016. People in the city of Santa Fe were more likely to vote this year than in similar elections in the past. Given that these areas are growing, new voters in these urban areas are going to outnumber new voters in rural areas. This will have a large impact in geographically and ideologically diverse districts like this one.

This presents a major problem for pollsters. Those who base their likely voter universes on past voting will miss these new voters. This clearly happened in the first congressional district in New Mexico where early polls missed the 10 point win by Deb Haaland by a wide margin (see here, here, and here). Whether Trump is producing a different Democratic electorate or other factors are at work, turning out new voters, especially in progressive/Democratic leaning areas, should be a high priority. Democrats lost in 2014 because 2012 Obama voters turned out at a lower rate than Romney voters. The 2018 electorate will be different. It will be up to the campaigns to effectively turn out Democratic leaning voters (and campaign on a message that resonates) who need less of a push than they did before Trump became president.