Support for impeaching President Donald Trump has fallen 7 points since December, a CNN Poll conducted by SSRS finds, following Speaker of the House Nancy Pelosi calling impeachment “so divisive to the country.”

The decline – from 43% in favor in December to 36% now – stems largely from a change in Democratic views on impeaching the President. In December, 80% of self-identified Democrats said they were in favor of impeachment – that now stands at 68%, a 12-point dip. Among independents and Republicans, support for impeachment has fallen 3 points over the same time.

The only major subgroup among which the decline was larger than among Democrats is college graduates: 50% backed impeachment in December, 35% do so now. Combining the two to look at Democrats with college degrees, support for impeachment fell 17 points from 79% in December to 62% now.

RELATED: Full poll results

Despite Pelosi’s comments, some members of the Democratic caucus in the House of Representatives have vowed to pursue impeachment this year.

Overall, the share of Americans backing impeachment for Trump is the lowest level found since CNN began asking about it last June. Support had been as high as 47% last fall.

Even though Americans are shying away from impeachment, only about 4 in 10 say Democrats in Congress are overreaching in the investigations of the President they have launched since taking control of the House in January. About a third (34%) say congressional Democrats are doing the right amount, and 22% feel they’re doing too little.

But Democrats in the House feel they aren’t getting sufficient cooperation from the President or his team. The top Democrat on the House Oversight Committee, Rep. Elijah Cummings of Maryland, wrote in a Washington Post op-ed published Tuesday that the Trump administration has not turned over “a single piece of paper to our committee or made a single official available for testimony during the 116th Congress.”

According to the poll, conducted before Cummings’ article, most Americans think the President isn’t doing enough to cooperate with Democratic investigations (53% too little) while 32% say it’s the right amount and 9% say the President is doing too much.

About two-thirds (67%) say the President should release his tax returns publicly, about the same as in October.

Special counsel Robert Mueller’s approval rating for handling the investigation into Russian interference in the presidential election in 2016 have rebounded – 48% now approve while 37% disapprove, a shift from near-even divides on this question in both early February and December polling. And 56% say they consider Russia’s efforts to influence the election a serious matter that should be fully investigated, while 38% consider it mainly an effort to discredit Trump’s presidency. Those results have been roughly steady over the last few months.

At the same time, there’s been a slight increase in the share of Americans who think Trump has told the truth publicly about the investigation (40% say so now, up slightly from 36% in December), though a majority (53%) still feel his statements have been mostly or completely false. The President’s approval rating for handling the investigation holds steady at 32%.

There’s also a more even divide now than in December over whether the investigation is likely to implicate Trump personally. In the new poll, 47% say it’s very or somewhat likely the president will be personally implicated in wrongdoing, while 48% say that’s not too or not at all likely. In December, 50% thought it was likely, 43% unlikely. That shift comes almost entirely among Republicans: In December, 23% thought it likely Trump would be implicated in wrongdoing; now, just 13% say the same.

Regardless of the outcome, the poll suggests Americans will be glad to support a public report on Mueller’s findings. Eighty seven percent say the findings should be made public, 9% say that’s not necessary.

The CNN Poll was conducted by SSRS March 14 through 17 among a random national sample of 1,003 adults reached on landlines or cellphones by a live interviewer. Results for the full sample have a margin of sampling error of plus or minus 3.8 percentage points; it is larger for subgroups.