Historically high negative perceptions of Donald Trump and Hillary Clinton could hurt the reliability of data used by campaigns up and down the ballot, political pollsters and analytics wonks say.
So political data firms including Deep Root Analytics, which worked for Jeb Bush's presidential campaign and is aiding Ohio Senator Rob Portman in his closely watched re-election effort, are now making adjustments to try to account for this election cycle's unusual paradigm.
Deep Root Analytics is investing a six-figure sum to revamp its data models to reflect the mindsets of today's voters, from free-trade Republicans who might support Ms. Clinton for president to Democrats seeking tariffs who could vote for Mr. Trump, and the impact such crossover voting would have on down-ballot candidates like Mr. Portman.
"We're due for a refresh of those audiences," said Brent McGoldrick, CEO of Deep Root Analytics, which develops data models to inform how political and commercial advertisers target TV ads. The company is also working with three other Senate candidates and several battleground House campaigns.
Political data and analytics professionals often suggest that using models based on voting in the last election is akin to fighting the last war. While firms such as Deep Root incorporate new information on a regular or semi-regular basis, however, they still tend to use models from previous elections as a foundation for the new ones.
This time around, Deep Root's data overhaul is different, according to the company. Deep Root recently completed a survey involving around 7,400 interviews with registered voters across the nation via landlines and mobile phones and is now integrating the survey results back into its voter file data. The process creates voter profiles that are used to build models of crossover voters, segment the electorate on a broader basis and prioritize coalitions of voters that will be important in certain races.
For instance, the company asked questions to determine voters' likelihood to support Democratic congressional candidates versus Ms. Clinton, the likelihood they'd vote for GOP congressional candidates versus Mr. Trump, and their stances on issues related to trade and foreign policy.
"The usual models don't apply this year," said Mr. McGoldrick, adding that conducting such a broad-scale field survey is rare in the middle of an election cycle, particularly in relation to TV targeting. "Down-ballot campaigns have to potentially target voters who might pull the lever for the other party at the top of the ticket," he said.
GOP technology and data consultancy Optimus has prepared to measure the impact of the party's nominee on down-ballot candidates for over a year by using multiple rounds of surveys and voter scores as benchmarks for current models.
"There are people who in the past have been reliably Republican but aren't necessarily going to vote Republican when they show up," said Scott Tranter, co-founder of Optimus. "There are certain things we are doing in the turnout modeling to adjust for the fact that there will be a different turnout profile."
Optimus recently conducted follow-up phone survey calls in swing states including Ohio and Florida to compare how voters stand today on Mr. Trump, who clinched the GOP presidental nomination on Thursday, with their feelings when Mr. Bush and others were still running against him. The findings can help illuminate challenges facing down-ballot candidates such as Mr. Portman, the Republican incumbent in his Senate race. He's facing Democratic Ohio Governor Ted Strickland.
If a Portman supporter in Ohio had a 90% chance of voting for Mr. Bush last summer but now has only a 60% chance of voting for Mr. Trump over Ms. Clinton, for example, that voter is far less likely to go to the polls than would normally be expected.
"That's a pretty staggering loss for the top of the ticket," said Mr. Tranter.
Though Democrats don't seem as worried about the down-ballot impact of Ms. Clinton, their presumptive presidental nominee, as Republicans are with Mr. Trump, they aren't ignoring the risk, either.
"There is a unique dynamic this year," said Dan Porter, co-founder and head of data science at Democratic analytics firm BlueLabs. However, he added, "It's standard practice for us to constantly validate and refresh our models throughout the cycle."
Republican research and polling firm Wilson Perkins Allen Opinion Research, which worked with Senator Ted Cruz's presidential campaign, inserts new questions into the surveys it conducts nationally and in select states on an ongoing basis to reflect voters' views as they morph throughout an election cycle. Yet the firm's chief research officer, Byron Allen, said the situation this cycle is calling on data firms to focus much earlier than usual on how presidential candidates will affect candidates for the Senate, the House, governorships and elsewhere.
"We will be a need to do a really big overhaul of our models so we can identify Trump Democrats and Clinton Republicans, and people who are voting in a different way than they typically would because of the candidates on the ballot," Mr. Allen said.
"Some chunk of Republican voters just may not show up," he added. "That's a bigger concern in some ways than trying to convince someone to split the ticket."