The Department of Education recently issued new guidelines on campus policies on sexual assault, including a directive that judicial bodies investigating sexual assault allegations employ the “preponderance of the evidence” standard in their deliberations.

“Preponderance of the evidence” is the standard commonly used in resolving civil cases — lawsuits — in the United States. It basically means that the question at hand will be resolved by a determination of which party’s version of events is more likely to be true. (“Beyond a reasonable doubt” is the standard of guilt used in criminal cases in the US, and there are other standards used in other circumstances, too.)

I don’t yet have an opinion on the DOE’s directive, and I think it’s an important question, so I’ve been reading up on it. And I just noticed something really weird.

As I noted above, “preponderance of the evidence” basically means that the judicial body will determine which side of the case, based on the evidence, is more likely to be in the right. If they come down on the side of the complainant, even hesitantly, the defendant is found guilty. It makes no difference, in other words, whether they’re completely convinced or have major doubts — whichever side they think is more likely to be in the right is the side that wins.

A common way of explaining this is to say that preponderance of the evidence means that if even 51% of the evidence presented supports one party, that party gets the decision. It’s an arbitrary number, of course — there’s nothing magical about 51% as opposed to 55% or 50.623% — but it gets the concept across. Whoever has the stronger evidence wins. Period.

So you see the 51% thing a lot. It’s all over the place — the number 51 appears in about a quarter of all web hits for the phrase “preponderance of the evidence.” But occasionally you see other numbers, like 50.1%, or even 50.000001%. And here’s where it gets interesting.

The number 50.1 appears only rarely in Google hits on “preponderance of the evidence” — a few times per thousand. But in pages in which the word “rape” or one of its variants appear as well, 50.1 shows up almost three times as often. You see a similar bump for 50.01, 50.001, etc.

Those numbers aren’t that big. It doesn’t seem to be a complete glitch — the numbers go in the opposite direction when you add “civil law” to the search instead of “rape,” for instance — but the magnitude isn’t huge.

Check out what happens when you substitute “campus sexual assault” for rape, though. The number 50.01 shows up three times as often as you’d expect, and 50.0001 shows up nearly seven times as often. Plug in “Title IX,” the campus sex-discrimination law on which the ruling was based, and 50.0001 shows up more than fifteen times as often as it should.

What does all this mean?

Well, one thing it doesn’t mean is that a campus assault case in which 50.0001% of the evidence supports the complainant is going to result in a conviction. The very concept of “50.0001% of the evidence” is meaningless — the idea that you could quantify the evidence in a sexual assault case to a precision of two parts in a million is absurd.

And that, of course, is why “50.0001%” is a figure of speech that you hardly ever see in discussions of this legal standard. It’s not coherent. It’s not meaningful. It’s not illuminating of the issues at stake.

And that is precisely why it keeps coming up in discussions of Title IX and campus sexual assault. Because “preponderance of the evidence means the side with 50.0001% of the evidence wins” is a lot scarier than “preponderance of the evidence means that the relevant body looks at all the evidence and rules for the party it thinks has the stronger case.”