Fairleigh Dickinson's PublicMind Polls released a doozy of a survey on Wednesday afternoon. According to their research, Americans generally want new gun laws. Oh, and a quarter of us see a conspiracy in the Sandy Hook shootings and three-in-ten think armed rebellion is imminently necessary.

PublicMind buries those second two datapoints in the third paragraph of their press release (which was spotted by Talking Points Memo). You'd think that research indicating that some 89 million Americans think armed rebellion might be necessary "in the next few years" would make it into the headline (we did!) or at least one of the first few lines. But, nope. (A bit of optimism: The survey does have a 3.4 percent margin of error, so it could be as few as 79 million Americans who are reaching for their AR-15s.)

Here's how the armed rebellion poll question breaks down by demographic group. Only two groups were more likely to agree with the need for rebellion than to disagree: Republicans and people who never went to college.

Please notice those gray sections of each bar. Those are the people who said they neither agreed nor disagreed with the question, or who said they weren't sure ("I am not sure if we will need armed rebellion before 2018"), or who refused to answer. So there's a lot of people who were a iffy on the question, which is some consolation. It also suggests a weirdness to the methodology — most polls have far, far fewer respondents who fall into that "other" category. (See, for example, the Washington Post poll we reported on Wednesday morning.)

The data on Sandy Hook — full question: "Some people are hiding the truth about the school shooting at Sandy Hook Elementary in order to advance a political agenda" — is similarly odd.

How do 37 percent of women neither agree nor disagree with that? How does that happen? It just doesn't make sense.

We're left with only one conclusion. The poll is at-best semi-scientific and should probably not be taken seriously. It certainly should not be written about by other media outlets. If you want to see the topline findings about gun laws, you're going to have to look somewhere else.

Update, 11:24 p.m.: Fairleigh Dickinson's Dan Cassino reached out to take issue with the representation of the poll as "semi-scientific."

This was a nationwide simple random sample telephone survey carried out by professional callers, including a cell-phone only oversample. The sample was then weighted according to known demographic (but not partisan) characteristics. Quite simply, I'm a scientist, and the end of all of these surveys is in peer-reviewed journals: the results are surprising, they're overstated in the media, but they're quite real, and quite scientific.

When pressed to explain the unusually high number of "neither" responses, he responded:

[T]hese are loaded issues: a lot of people don't want to take a stand on them if they can avoid it, and on sensitive issues like this, it's borderline unethical for us to push people into coming down one way or the other if they don't want to.

This methodology and the results are not typical.