Just as journalism is the first draft of history, exit polls let us take some first stabs at understanding public opinion. On the matter of politics and race, Donald Trump’s victory will provide plenty of fodder for opinion columns, in-depth survey analyses, fieldwork and historical studies.

Given the various studies already conducted, factors such as racial resentment and immigration conservatism will likely play important roles in explaining white voting behavior, in addition to factors such as economic anxiety and distrust of political elites.

Our understanding of race and presidential voting, however, also needs to look comparatively across groups. Looking at the exit poll data from the 2016 election, and comparing them over time, the story on race and presidential voting gets a bit more complicated.

One caveat about the national exit polls: They are designed to broadly represent the American electorate. Their sample design and lack of Asian language support may not be accurate in estimating the absolute support of subgroups in a given election. But the comparisons across years should give us a decent sense of how those groups move from year to year.

It’s also important to note that polling firms Latino Decisions and Asian American Decisions found much higher support for Clinton among Latinos and Asian American voters in a survey fielded in the week prior to Election Day. As I’ve noted before, it’s possible that differential turnout on Election Day among Democrats and Republicans might account for part of this discrepancy.

There are different benchmarks one can use to see how Trump performed compared with prior Republican nominees. Comparing yesterday’s results with 2012, as this Washington Post feature does, shows that Trump actually performed slightly worse among white voters than Mitt Romney did. He did, however, perform better than Romney among blacks, Latinos and Asian Americans, making it more difficult to claim that racial resentment was the dominant factor explaining Trump’s support nationally.

The 2012 election was not an open one, however, since Obama was running for reelection. When we compare Republican nominee performances in prior open elections, we see that Trump made important gains among black voters as well as whites. The figure below presents net Republican support, meaning voter support for the Republican nominee minus voter support for the Democratic nominee. The GOP advantage among white voters was 9 points higher in 2016 than in 2008, which may be due as much to whites’ reluctance to vote for Obama in 2008 as to their enthusiasm for Trump in 2016.

At the same time, Trump performed better with black voters than McCain did in 2008, and on par with Bush’s performance in 2000.

Trump performed as well as McCain did with Latino voters in 2008, and did only somewhat worse than Bush in 2000 — which is surprising given the way that Trump launched his campaign with incendiary remarks about Mexican rapists, kicked out a prominent Mexican American journalist from a news conference, and questioned the fairness of Hispanic judges.

Finally, among Asian Americans, Trump performed significantly worse than Bush did in 2000, and marginally worse than McCain did in 2008.

How do we explain these time trends? Why were many Latino voters apparently willing to overlook Trump’s string of disrespectful remarks, including ones that House Speaker Paul D. Ryan characterized as “the textbook definition” of racism? Asking this very question might cast doubt on the exit poll findings about Latinos. But we should also remember that Clinton also failed to win the majority of white women, despite Trump’s many negative remarks, ranging from Megyn Kelly to the Access Hollywood tape to various assault allegations, all of which got wide news coverage.

Absent final turnout numbers, it is still too early to assess whether these shifting vote patterns are the result of differential turnout among Clinton and Trump supporters or the result of genuine voter conversion. It’s possible that a sizable chunk of Latino Clinton supporters, in addition to white women, African Americans, and Asian Americans, did not show up on Election Day. It’s also possible that a significant portion of these voters were willing to overlook Trump’s incendiary remarks and vote for him based on other factors, like the need to shake up “politics as usual.”

Time will tell how well the national exit polls reflect the vote choices of women, Latinos, African Americans, and Asian Americans. Still, comparisons of the national exit polls over time reveal some important, and potentially unsettling, questions about what we presume to be true about voting among America’s racial minorities. Did Hillary Clinton have an even tougher road with African Americans than Al Gore did in 2000, or did she benefit from multiple campaign appearances by Barack and Michelle Obama? Is immigration still a litmus test for Latino voters? And can we expect Asian Americans to shift back toward the Republican Party in future elections, or is their party identification hardening and making them more immune to big shifts in candidate choice?

This week’s exit polls, in light of prior elections, raise these important questions about race and public opinion — not just about whites, but also about blacks, Latinos and Asian Americans. And it will be up to a new generation of research — done not just through surveys, but also through case studies and community fieldwork — to help us better understand what happened this week, and what it will mean for the future of U.S. politics.

Karthick Ramakrishnan is professor of public policy and political science at the University of California, Riverside, and author, most recently, of “Framing Immigrants: News Coverage, Public Opinion and Policy.” Find him on Twitter @karthickr.