Did Friday’s tape of Donald Trump bragging about his celebrity status and ability to “grab them by the p****” change a lot of minds? It depends on which poll one reads –and even which NBC poll. Allahpundit linked the toplines of a new NBC/Wall Street Journal poll that shows Trump trailing Hillary Clinton by eleven points (14 head-to-head), but that has a few significant caveats to consider.

The Wall Street Journal/NBC News survey was taken Saturday and Sunday, after a recording became public on Friday of Mr. Trump making what he thought were private comments about women—but before he defended himself in Sunday’s televised debate with Hillary Clinton. In that debate, Mr. Trump said he had been engaging in “locker room talk’’ and revived accusations of sexual assault by Mrs. Clinton’s husband, former President Bill Clinton.

In the new survey, Mrs. Clinton jumped to an 11-point lead over Mr. Trump among likely voters on a ballot including third-party candidates, up from 6 percentage points in September.

Interviews for the poll were continuing Monday and could signal whether Mr. Trump had rebounded after Sunday’s debate.

That’s a big crater to appear just four weeks out from election day. However, three caveats become immediately apparent. First, the poll uses a small sample for a national election poll, which Nate Silver pointed out almost immediately on Twitter. The previous NBC/WSJ poll had almost twice as many likely voters in its sample (922, three weeks ago). It’s literally the smallest sample since Trump clinched the nomination, according to RCP’s aggregation for four-way polling.

Second caveat: Why is it so small? As the WSJ report notes, the polling is still ongoing. What we’re seeing is rough data from the midpoint of the poll. The surveys completed at this point might not be demographically balanced well enough to rely on this result. It’s a bit like reading early exit polling.

Third caveat: The polling might be too close to the impact event, and not enough to encompass its overall context. Saturday might have gotten the biggest reaction, but the following days could have seen a return to the race’s status quo ante. Plus, without the debate reaction, this interim result isn’t really reflective of the race as it stands now.

That’s not to say that we can entirely discount this poll, either. The tape did have some impact, according to the NBC/Survey Monkey tracking poll data from the weekend, but perhaps not as much as one would have expected:

After a 2005 recording of Donald Trump making lewd comments about women emerged on Friday, 63 percent of likely voters said they do not think the Republican nominee respects women, according to the latest NBC News|SurveyMonkey Weekly Election Tracking Poll. In the four days the poll was in the field prior to the revelation of Trump’s comments on Friday, 55 percent of likely voters said they did not think he respects women.

The damage done takes a surprising turn:

Before the recording was made public, 64 percent of likely women voters said they did not think Trump respects women. After the recording was made public, 69 percent of women said they did not think Trump respects women, including 53 percent who said he does not respect women at all.

Though a large number of women voters have favored Clinton throughout the campaign—especially in the wake of the first debate—more men have generally favored Trump. After the video surfaced Friday, however, men’s attitudes regarding Trump and women swung even more than it did among women.

Prior to the recording’s release on Friday, 45 percent of likely men voters said they do not think Trump respects women. After the recording was made public, however, that number jumped 10 points to 55 percent. Men have been a core constituency for Trump, and his 2005 statements have affected their view of how the Republican nominee treats women significantly.

That’s only a five-point shift among women, which might not make a big difference, but a ten-point swing among men, which might — if it lasts. It does tend to negate the Team Trump spin that this was just “locker-room talk” typical of men; apparently, most men don’t experience that kind of conversation after all. However, those men probably won’t cast their vote on the basis of that character trait, and they’re likely to stick with Trump.

The overall result in the poll this morning — Hillary Clinton up 46/41 in the four-way race — tends to bear that out. The chart of the toplines of the poll barely budged from the week before, and the slight change that did take place went in Trump’s favor. Sometimes these big events need more time to impact a race, but at least this far it seems that this episode might matter less than many assumed a couple of days ago.

Among independents, that seems to be particularly the case. Almost two-thirds of unaffiliated likely voters (63%) say the tape makes no difference in their voting decision this year. Three in ten say it will make them less likely to vote for Trump, but as Aaron Blake points out, that might mean even less than it seems:

Views of Donald Trump were already dim before The Washington Post published a video Friday showing him talking in very lewd and sexually aggressive terms about women in 2005. And the video appears to have done more damage, according to a new poll.

At the same time, the survey of the independent voters who are likely to determine this presidential race overwhelmingly say the video makes no difference to them.

So more than twice as many say it’s not a negative for them. Given the wall-to-wall coverage of the video and the large movement within the GOP to get Trump to drop out, that doesn’t seem like a lot. The 31 percent could simply be the many Democratic-leaning independents who already were going to vote against him, after all.

Again, it’s almost certainly too soon to see the full impact of the video, but Trump’s bounceback in last night’s debate might mitigate that, too. It might be that voters are more concerned about issues and dysfunction that a decade-old tape of Trump engaging in disgusting braggadocio isn’t really what moves the needle.

However, even if that’s true, it’s still a problem for Trump. In order to win in four weeks, Trump has to convince voters to switch to his side and trust him over Hillary Clinton. It’s not enough to simply hold serve with one’s own base in a general election. Even if this episode does minimal damage, the opportunity costs for losing precious time and credibility may be significant enough to lose the election, especially since Trump has been fighting from behind nearly the entire way since the convention.

Either way, the polling on this won’t really be reliable until well after Thursday’s debate — assuming we don’t get any other bombshell oppo dumps on Trump. That’s not a bet I’d put money on.