The past few weeks have been hard on the Clinton campaign and it’s showing up in the polls as an erosion of support nationwide. But only in the past few days have I started receiving panicky email from liberal and #NeverTrump readers finding themselves for the first time imagining a Trump presidency as more than a reality TV plot turn. Despite my efforts to caution against reading too much into individual polling, two surveys released yesterday showing Trump with a low single-digit lead in Ohio seem to have breached a psychological barrier for some who had taken comfort in the blue bulwark which for months had dominated the electoral map. So allow me to offer some perspective on the Ohio polls in an effort to reassure those readers, then address the important message the polls are giving us.

About a week ago, I wrote that the structure of the presidential race had not changed in a meaningful way for months. In the interim, Hillary Clinton developed pneumonia, was criticized for keeping it secret for two days, was captured on video losing her balance as she left a September 11 remembrance in New York, faced a wall of critical coverage for packing half of Trump’s supporters into a “basket of deplorables” (an awkward turn of phrase which brings to mind something you should be able to order on Amazon), then was invisible in the campaign as she recovered from her illness. And the structure of the presidential race still has not changed in a meaningful way. As we have seen at other moments when the news agenda focused negatively on Clinton, weak supporters who may not particularly like her headed to Gary Johnson or undecided, but Clinton and Trump continued to bounce around in the same range of support.

So what happened in Ohio? Four polls were released covering roughly the same survey period: a Bloomberg Politics poll showing Trump up by five, a CNN poll showing Trump up by four, a Quinnipiac poll showing Trump up by one, and a CBS/YouGov poll showing Clinton up by seven. Clearly all these results cannot be true, even if you account for the polls’ margin of error, just as it is unlikely that a contemporaneous Quinnipiac national poll showing Clinton up by five exists in the same universe as their Ohio poll, or a contemporaneous CBS national poll showing Clinton up by two exists in the same universe as their Ohio poll. Ohio tracks closely with national opinion—that’s why it’s a swing state. If Clinton is leading by five nationally she is almost certainly not down by one in Ohio, and if she is leading by two nationally she may be ahead in Ohio, but not by seven. Do you see why I caution against reading too much into individual polls?

It can be tempting for partisans to dismiss results they don’t like by digging into the details to look for assumptions that don’t make sense. Both sides do it and it is a waste of time. But the details of a poll can tell us useful things about the behavior of the electorate. The Bloomberg Ohio poll consists of a sample that’s weighted heavily toward groups favorable to Donald Trump. Republicans enjoy a seven point advantage and 55% of respondents are over age 50. Fifty-seven percent do not have a college degree and 83% are white. The CNN sample has a closer partisan split but skews even more heavily toward older voters. It is not difficult to imagine that an older, whiter, less well educated electorate would favor Trump. These are his core groups. So while these polls may not be able to tell us if the Ohio electorate will look like this in November, they can point us toward the likely outcome if it does.

It is possible that Democratic constituencies were underrepresented in these surveys because they were less motivated to answer the phone during a particularly demoralizing phase of the campaign for Clinton. But it would be a mistake to dismiss the results on these grounds, because demoralized voters who don’t answer surveys are also less likely to vote. Now there is reason to believe Hillary can energize her supporters by returning to the campaign healthy and engaged, and the inevitability that Trump will again focus attention on himself should give her plenty of opportunities to go on offense. She has the advantage of an elaborate ground game to boost her vote among weak supporters who might not otherwise turn out, and—for those readers still seeking reassurance—she can win the presidency without Ohio but Trump cannot.

These polls should remind us of one of the fundamentals of this contest: Trump’s voters are more enthusiastic but Clinton’s voters are more numerous. Trump can’t win if the electorate resembles 2008 or 2012, and Clinton has a little under two months to make sure it does.