So, let's talk about confidence levels. Remember that we had this measure of central tendency, that was our mean, for instance. The mean of a single group, the means of the difference between two groups. But that is a single value that represents all of the other data values. Now we need to infer that number, that's a sample statistic to the population, to a population parameter. Now, the measure of central tendency, we also call it a point estimate, but that's all it is, it really is just an estimate. It really doesn't give us an indication of how spread out that data was. Now, we know measures of spread, we could use standard deviation, so could we use the values for one standard deviation away from the mean, or the difference means, to say well, we think the patient or the population parameter is within those values. Now the problem there is really, we cannot. Because remember, the sample data that we have can be quite skewed. And working out a standard deviation of that is almost never going to be a good idea to infer our result to a population parameter. We need something different. We need some lower bound and upper bound, and we need some level of assurance that we think that the patient or the population parameter should be within those upper bound and lower bound. Now how confident are we? Those are the confidence levels. How confident are we that the population parameter falls within those limits that we are going to have. Well, think about it. If we chose that it's possible that the population parameter can fall anywhere between negative infinity and positive infinity, then we could be 100% confident that the value of the true population is going forward in those values. But we can't choose negative and positive infinity, we've got to be realistic about it. But the closer we get and the smaller we make those bins, the larger the probability is that the true population parameter falls outside of those. So we are no longer 100% confident that the population parameter falls within those lower and upper bound limits that we are going to have. Now it is customary to calculate 95% confidence intervals. And we would say, look, probably the true population parameter falls within those limits. Now that's not quite correct. I just want to say, it's not quite correct. And we'll look at the exact definition or the correct exact explanation in the next section. But we could say we are around 95% confident. Those are confidence levels of the true population parameter falling between those limits that we do give. Now I use the word probability. That's our old friend, yes, the central limit theorem. We can really construct this same sort of graph that we had, that same sort of plot with the central limit theorem, and work out an area under the curve for which we would find 95% of all the values. So, really our old friend, nothing different from what we've done before. At the center though, now would be the mean of one group of patients, or the difference between the means of two groups. That would be the mean and around, that means we could construct a 95% cutoff values, for which 95% of the values would fall between, and that's exactly our level of confidence. In the next section we're going to discuss actual confidence intervals, and we'll get to the true meaning of what that 95% represents.