Looking at reliable recent polls, you could come away with two contradictory conclusions: Donald Trump is cratering, allowing Hillary Clinton to run away with the presidential race. Or Trump has survived an awful month and is surprisingly competitive.

I'm going presume to tell you what the state of play really is by looking at multiple surveys and extrapolating a bit.

Clinton, though she remains an unpopular candidate, has an advantage of about 7 points, though it's slightly less when third- and fourth-party candidates are included. State surveys that show Trump running almost even in battleground states are exaggerated. More worrisome for Republicans are the internal dynamics of these findings that suggest deep trouble for the presumptive nominee and perhaps the party.

Start with gold standard polls. Any list of best pollsters, from the data guru Nate Silver to political practitioners, would include Ann Selzer, the Bloomberg Politics pollster, Peter Hart and Bill McInturff, who conduct the Wall Street Journal/NBC News poll, and Gary Langer, who does surveys for ABC News/Washington Post survey.

In national polls this month, Bloomberg and the Post/ABC gave Clinton a 12-point advantage, close to a landslide, while the NBC/Wall Street Journal found a less-comfortable 5-point edge.

There's a fairly simple explanation: the stated political affiliation of the respondents. The ABC/Post poll had more Democrats, by 12 percentage points. Democrats had a 7-point advantage among likely voters in the Bloomberg survey and a 4-point upper hand in the NBC/Wall Street Journal poll.

In 2012, according to Election Day exit polls, Democrats had a 6-point edge; four years earlier, it was 7 points.

Pollsters often will weight or slightly adjust final data for gender or race. But the best ones do not do that for political affiliation; this is subjective and reflects voters' attitudes at the moment they are called. In 2012, 92 percent of Democrats voted for President Barack Obama; 93 percent of Republicans voted for Romney. In Bloomberg's latest poll, 93 percent of Democrats said they would vote for Clinton; 84 percent of Republicans opted for Trump.

On the state level, Republicans are citing a CBS survey over the weekend that showed a very close race in battleground states, a 1-point Clinton advantage in Democratic-leaning Colorado. They also note that Real Clear Politics, which aggregates data, suggests that her lead in Pennsylvania -- a graveyard for Republican presidential hopes -- is less than 1 point.

There are a few problems here. The Colorado poll had a high number of undecided voters.  And the Pennsylvania average is based on two polls, one a bit dubious.

Here's a rule of thumb in considering battleground surveys: If the national needle moves one way or the other so will these states. 

To be sure, in isolation, Clinton's numbers aren't impressive. In the Wall Street Journal/NBC poll, she trails Trump on handling the economy and terrorism and gets trounced on the question of bringing change to Washington.

But the numbers for Trump and Republicans are considerably worse; in the Post/ABC poll, he gets a 70 percent unfavorable rating and 64 percent of respondents said he's unqualified to be president. Bloomberg Politics finds that 55 percent of likely voters said they would not vote for him under any circumstances. His ceiling appears to be close to his floor.

And he seems to be hurting the Republican brand. In the Bloomberg Politics survey, by almost a 2-to-1 margin, voters were negative about Republicans, the lowest number in recent memory.

Some Republicans argue that they will outperform the polls as Trump's backers are more committed and enthusiastic thanClinton's, who engenders much less enthusiasm than Obama did. Democrats counter that the general election profile is more favorable to them than the past couple elections, with more people of color.

Ken Goldstein, a polling expert and Bloomberg Politics consultant, sees early evidence that Clinton and Democrats do a little better with likely voters than when all registered voters are surveyed. He cites a Wisconsin poll that showed both Clinton and Senate candidate Russ Feingold opening up a larger lead when the contest is narrowed down to likely voters.

"Republican identifiers who have an unfavorable view of Trump are much less likely to say they're going to vote," Goldstein says. "This is a huge problem for down-ballot Republicans."