A woman poses with a cutout of Democratic presidential candidate Bernie Sanders (Vt.) before a campaign rally at Cass Technical High School in Detroit on Oct. 27. (Jeff Kowalsky/AFP)

There’s a sort of trickle-down hierarchy of questions worth asking to understand the 2020 election.

It starts with the states: Will 2020 come down to the same Midwestern and Rust Belt states that it did last time — Wisconsin, Michigan and Pennsylvania?

We go down a level: If so, why did President Trump win those three states in 2016? Why didn’t he win Minnesota, a state that is similar to those three states but where he came up short by a small margin? Did it come down to swing voters deciding to give him a shot?

Down another level: If so, how are those same voters likely to vote next year?

It’s easy to see how that question is both potentially important and, potentially, unimportant to the outcome of next year’s presidential election. If Trump ends up winning Pennsylvania by 10 points or losing Wisconsin by 20, small shifts in opinion among voters in those states probably weren’t what made the difference.

It’s safe to assume, though, that answering those lower-level questions will, in fact, inform our expectations. A new poll from the Kaiser Family Foundation and Cook Political Report released Thursday offers insights that might help us do so.

The pollsters didn’t ask people to choose between Trump and prospective Democratic candidates by name. Instead, likely support was measured more abstractly, asking respondents if they were definitely planning to vote for Trump, definitely planning to vote for the Democratic nominee or if they fell somewhere in between. Across the four states included in the poll (those mentioned above), about 40 percent of respondents didn’t have a “definite” preference in next year’s election. About 20 percent indicated that they would probably support Trump or the Democrat, with the other 20 percent identifying themselves as undecided.

Across all four states, the distribution of those responses was fairly even.

(Philip Bump/The Washington Post)

The 40 percent of respondents who weren’t definite in their intentions differed in two key aspects from respondents who indicated that their minds were made up. More than half of those who had already made up their minds identified as Democrats or Democrat-leaning independents; a plurality of those who hadn’t made up their minds identified as Republicans or Republican-leaning.

Most of those who hadn’t made up their minds identified themselves as moderates.

(Philip Bump/The Washington Post)

In other words, most of those undecided respondents are moderate, battleground-state swing voters — a combination of words that will trigger a salivation reflex in any pollster. Purely undecided voters, those not leaning in one direction or the other, make up about a fifth of respondents in each state, more than enough to swing results. Understanding how they view the candidates, then, would help up answer our questions about 2020.

KFF and Cook made doing so fairly easy by presenting respondents with questions focused on policy proposals introduced by Democratic candidates. Here’s how they were viewed overall:

Issue Good idea Bad idea
A pathway to citizenship for immigrants who are in the U.S. illegally 70% 28
A Green New Deal that would address climate change through new regulations and increases in government spending on green jobs and energy efficient infrastructure 67 31
A ban on future sales of assault weapons and military-style rifles, like the AR-15 66 33
A ban on ownership of assault weapons and military-style rifles, like the AR-15, including a mandatory buyback program for current owners 54 45
A ban on hydraulic fracturing, or fracking, a drilling method that uses high pressure water and chemicals to extract natural gas and oil 40 54
A national Medicare-for-all plan that would eliminate private health insurance 36 62
No longer detaining people for crossing the U.S. border illegally 27 71

That’s interesting in its own right, but we’re interested in a specific subset of responses to those policies: how they’re viewed by those who aren’t yet committed to a candidate next year. KFF and Cook provided us with a breakdown of responses by category.

On net, here’s how each group viewed the policies.

(Philip Bump/The Washington Post)

A majority of those undecided in 2020 supported the first four policies, including introducing a pathway to citizenship, the Green New Deal and bans on military-style rifles — including one with a mandatory buyback. About 7 in 10 undecided respondents said they thought the first two policies were a good idea; on net, the first three were preferred by undecided voters by about 2 to 1.

The last three — fracking, Medicare-for-all with the elimination of private insurance and decriminalizing illegal border crossings — were viewed as bad ideas by more than half of undecided respondents. Only on the least popular proposal (border crossings) did less than a third of undecided respondents say that the proposal was a good idea.

This very specific question about the views of undecided battleground voters is of enormous interest to Democratic presidential campaigns and observers. Many Democrats worry that a policy platform that’s too liberal will turn away voters like that. The KFF-Cook poll certainly does hint at danger zones, like an expansion of Medicare that eliminates private insurance, but even in that case, 40 percent of undecided respondents like the idea.

In other words, if 2020 comes down to these states and these swing voters and these issues presented in this way, good news for a Democratic nominee who maintains the same level of support that an unnamed, theoretical one does! If any part of that sentence is off the mark, though, things will get a bit murkier.