Another day, another potentially concerning poll from the New York Times. Two weeks ago, the New York Times/Siena poll showed a minimal debate bounce and a lead in Pennsylvania. Last week, they had Harris down in Arizona, Georgia, and North Carolina. And then, on Sunday, the New York Times released another series of polls showing deadlocked races in Wisconsin and Michigan, a deficit in Ohio, and a lead in the critical second congressional district of Nebraska. According to the poll, Harris leads 49-47 in Wisconsin, 48-47 in Michigan, and 52-43 in the second congressional district of Nebraska. Trump leads 50-44 in Ohio.

Once again, my phone filled up with messages from democratic activists, friends, and Message Box subscribers concerned about the poll. There are two buckets of reactions. The first is typical — the New York Times is biased against Kamala Harris and/or making the race seem close to drive traffic and revenue. The second is fear and exasperation that the race is so nauseatingly close.

I get it.

Because this poll drives the political discourse, and affects the emotional health of so many Democrats, I thought it would be worth addressing these concerns.

1. Should We Dismiss this Poll?

The New York Times/Siena poll is considered one of the best and most accurate. With the stipulation that no poll is perfect, I take their polls at face value because of their track record; and Nate Cohn, who runs the poll, is very transparent about their methodology and any possible flaws in their sample.

Now, that doesn’t mean these polls are all correct; I do not believe there has been a ten-point swing in Arizona or Georgia in about a month. Harris is unlikely to be up four in Pennsylvania and only one nationally, as their polling two weeks ago found. Most polling suggests that Wisconsin and Michigan are slightly easier— and Democratic operatives agree. Therefore, Harris being up four in Pennsylvania, only one in Michigan, and two in Wisconsin is a little surprising. Some folks dug deep into the crosstabs to find huge, unlikely shifts or other potential flaws. I would caution everyone that the sample sizes of these subgroups are small, and therefore, the margin of error is huge.