If you are my friend on Facebook, you saw my (admittedly stupid) status update: I am starting a paper. I’m going to do this paper EARLY. Seriously.

I was really planning on it, too. Seriously.

But then I started my bio lab.

And I am very sidetracked by my bio lab. We are supposed to do a chi-square analysis of some ant data. For anyone who has spent any amount of time in public health, or with population (ANY population – ants, monkeys, people)-based studies, a chi-square is Not A Big Deal. At ALL. But here is the statement, in the middle of the paper:

“Do all the calculations, refer to [Table] for the critical value, and decide if you will accept or reject the null hypothesis.”

Emphasis mine.

Anyone see a problem with this particular phrasing?!? I DO! So here it is: My major contribution to the collective knowledge of the internet. Here you go. Ready?


Let me try it again.

You can NEVER EVER EVER EVER EVER accept the null hypothesis.

You can reject the null hypothesis. You can fail to reject the null hypothesis. But you canNOT ACCEPT the null.

I can see you thinking: oh, but Rachel. That is totally an issue of semantics.

Actually, my friends, no, no, it is not.

I will try to explain.

The null states that the number of ants infected in population A are the same as the number of infected ants in population B. Get it? The two populations are the same with respect to the number of ants infected IN THIS SAMPLE OF ANTS. (That is an important distinction.)

The alternative hypothesis is this: The number of ants in population A are DIFFERENT from the number of ants in population B. Meaning that there are more, or less, infected ants in one population as compared to another.

Now, if we find there is an important distinction between the ant populations, we will reject the null hypothesis. We will say, “NO! One species of ant is way more infected than the other species.”

However, what if we find they are the same? We will say, “We cannot reject the null hypothesis. We do not have the evidence to suggest that it is any other way than they are the same”. We cannot say “We believe that all ant populations are the same with respect to infection.” We are simply saying, “we cannot conclude with any evidence that the null hypothesis is WRONG.”

Does that make sense?

If it doesn’t, just go with me on this one: You cannot prove a null hypothesis. You cannot accept a null hypothesis. All you can do is REJECT it, or fail to reject it. That’s it.

This is much more important in humans than in ant populations, but it’s statistics and thankfully, statistics is actually the same across disciplines. I know. I’ve gone through a few in the last 10 years.

If you would like more information, you are welcome to contact me.

I am now done with my Stats 101 post. Phew. That feels better now (a little bit.  Except I am still annoyed.)