The way to win a general election used to be pretty straightforward. There was a primary between several members of the party in which more centrist candidates would get pulled to the partisan pole. But, the party hoped, not too much — allowing the victor in the primary to still appeal to the political center in November. A successful candidate who could win a general election fell somewhere in the area of the solid line below.

Why? Because the political middle was where most voters were. Democrats and Republicans differed ideologically on a range of positions, but still agreed on a number of points. The Pew Research Center has asked a battery of ideological questions for several decades and, in the era of Bill Clinton, the partisan distribution looked like this.

If you’re Democratic candidate D1, you’re going to have a much better shot at picking up votes from Republicans than candidate D2. Same goes for R1 — more Democrats would be likely to consider R1 as a candidate than R2. In 1994, Pew found, 36 percent of Republicans were more liberal than the median Democrat, and 30 percent of Democrats were more conservative than the median Republican.

But that was in 1994, when Rep. Nancy Pelosi (D-Calif.) — now House minority leader — was seven years into her tenure and Sen. Charles E. Schumer (D-N.Y.) — now Senate minority leader — had been on the Hill for over a decade.

The most recent Pew data shows a much different distribution.

Suddenly, the calculus changes. The four candidates we indicated on the 1994 chart are still here, in the same positions. But D1 and R1 are no longer the obvious consensus candidates. For the Democrats, at least, D2 seems like the obviously better candidate, able to earn much more support from the far-left of the party than D1.

The presidential candidate in 2016 who advocated most vocally for how this dynamic had shifted wasn’t the eventual winner of either party’s nomination. It was Sen. Ted Cruz (R-Tex.), who pitched his candidacy on the idea that a conservative Republican — as opposed to a more middle-of-the-road candidate such as Mitt Romney or John McCain — would energize the broad hard-right wing of his party and increase turnout to take on the Democrats. Cruz’s political math wasn’t wrong; he just assumed that the wrong candidate would win and that the wrong political ideology would mobilize the voters. It was Donald Trump’s blend of far-far-right politics and more mainstream conservativism that earned him enough support to squeak by Hillary Clinton in the requisite states and become president.

He was aided by a related phenomenon: partisan loyalty. Pew has also found that hostility toward the other party has increased over time. In 2016, 58 percent of Republicans said they viewed Democrats very unfavorably, and 55 percent of Democrats said the same of Republicans. Those figures were up from 21 percent and 17 percent in 1994, respectively.

The effect has been fewer crossover votes from members of one party for candidates of the other. In 2016, not a single state that voted for a presidential candidate from one party elected a Senate candidate of the other. Over time, not only has the number of split-ticket Senate voters declined, so has the the deviation between presidential results and Senate results.

So Republicans in 2016 voted for Trump, both because they were skeptical of Clinton and because, well, they’re Republicans.

Those two trends — partisan loyalty and the polarization of political ideology — complicate the tactical question that Democrats face in 2018. Stoking the base’s white-hot anger at President Trump might boost turnout (as we’ve seen turnout increase in primaries and special elections, spurred by frustration at Trump). But might it also stoke Republicans to vote more heavily, too? Is a platform focused on economic and health-care issues a better strategy for success in 2018?

The last question is fairly easy to answer, based on the patterns above. There are certainly places and races in which centrist candidates will be battling to capture a majority of votes from an ideologically mixed electorate. But in many places, the distribution of votes will look more like the 2017 chart above, and candidates will be successful in making more partisan pitches, relying on partisan loyalty and a more-polarized party base to generate votes in November. A lot of Democrats, in other words, will use the Ted Cruz strategy: Run left and energize the left to actually come out and vote, boosting turnout for their side.

Turnout is a central issue in midterms, which overlaps with the questions raised above about focusing on Trump.

We noted last week that Trump’s base is more energized to vote for candidates as a show of support for him than we’ve seen in other recent midterm elections.

A recent article in the New York Times argued that opposition to Trump and Trump’s policies has driven a burst of support for Trump from otherwise-lukewarm Republicans.

But Republicans often have a turnout advantage in midterm elections, given that midterm elections get lower turnout than presidential races and given that the most frequent voters are older and wealthier — demographics that tend to vote more heavily Republican. It’s possible that the infrequent voters who came out in 2016 to support Trump might be similarly inspired to turn out to vote for his party this year, giving the Republicans an additional boost. But that was the operating theory before 2010, too: that infrequent voters who came out for Barack Obama in 2008 would vote for Democrats two years later. That wave didn’t materialize — and it’s not clear that Trump’s will, either.

Meanwhile, Democratic turnout is up consistently in primaries this year, certainly in part a function of Trump. If opposing Trump demonstrably spurs less-frequently-voting Democrats to vote in November and only might help spur infrequent Trump voters to the polls, it’s not a tough call for Democratic strategists.

So why the pushback on challenging Trump within the party? Because focusing on more partisan political positions is a fairly recently emerging factor in national politics. The Republican base has been pretty conservative for a while, but Democrats have only recently gotten more liberal.

The Democratic Party that elected Pelosi and Schumer doesn’t look like the Democratic Party that’s going to the polls this year, both literally (that is, demographically) and ideologically. Politicians are superstitious, preferring to rely on the strategy that they know worked than one that might not. Pelosi and Schumer are savvy enough to be able to test wind direction, certainly, but there’s a difference between knowing how the weather is changing and preparing for the storm.

My colleague Aaron Blake notes that Democrats don’t want a Democratic majority in the House to be overly focused on Trump. That’s a January problem. They need to get past the November problem first, and the sea change in American politics suggests that partisanship can be a path to political victory.