So, let's consider this alternative hypothesis. It really comes in two flavors. Let's use a hypothetical example. Imagine I took patients with acute appendicitis and I look at the admission white cell count and I divide them into two groups. One group is HIV positive, one group is HIV negative. I want to know is there a difference in the admission white cell count between those two groups of patients with acute appendicitis. Now think about it, my null hypothesis has got to state, no there's no difference in the means between those two patient groups. There's no difference in the means. My alternate hypothesis, I could state that in a variety of ways. I could just say, well, I think one will have, one group will have a mean white cell count more than the other or less than the other. That would be one way to state it, or I could say I don't know if one's going to be more or less than the other. They're just going be different, and that's two fundamental different ways to look at it. Now those two choices, either more than or less than, or just different, those we refer to as one tail and two tail test. One tail and two tail T test, for instance. Now look at this graph. This is what we refer to as a two tailed test. That is when our alternate hypothesis just states that there is going to be a difference between the groups. Now we find a difference in means. It will be a certain standard error away from the mean. But the alternative hypothesis was that they're just going to be different. So if turned out one was more than the other, or one was less than the other, you've got to do it on both sides of this graph. So you're actually multiplying by two under the curve because that difference, will be so many standard errors from the mean. You've got to draw that line on the graph. But, you've got to duplicate it on the other side, because the alternative hypothesis was that there's just a difference. So your area under the curve for that finding is going to be twice as large. Look at the next example. If we had a one-tailed hypothesis and this example we just said that one group was going have a value more than the other group. Now for us to have an area under the curve of 0.05. The computer calculates for us how many standard errors away from the mean, that would be, it draws that line. But it represents all of the 5%. And you can well imagine that that line could be slightly closer to the mean. If we now find a difference between the two groups and we convert that difference into units of standard error, and we plot that on the graph. It has a high likelihood of falling within that range because that's a bit bigger now on that side. We've lumped all of the 5% on the one side. It's easier to find a value that will fall into that 5%. We don't duplicate it on the other side. We don't multiply that by two because the alternative hypothesis was stated that this group is going to have a value less than or more than, in this instance, the other group. So, there's a fundamental difference in the p-value, purely by the statement of the alternative hypothesis. Now, in many of these such papers, you won't see, you won't find that in the method section was mentioned, whether these were one-tail or Two-tail test, so watch out for those type of articles. They proper. They tell you whether it was a one-tail or a two-tail p-test. Now, what should it be? Is there an absolute rule when to choose a one-tail test and when to choose a two-tail test? No, it's not. It clearly depends on the situation, the actual topic that has been addressed through that research article. It should be clear, logically clear, that you could choose a one-tailed test before the data collection and the analysis start, or whether it should really have been a two-tailed test. So, you really want to know that from the method section of any paper.