In this section, I'm going to discuss the very exciting topic of Confidence Intervals. You will see them as CI. Now, they are everywhere in the literature, you will see them mentioned after means, after standard deviations, after variances, so many 95% usually CI confidence interval. There are many uncertainties about what are the means, what are means exactly and I think there's probably a poor understanding of the term Confidence Intervals. Now remember what our aim was, our aim was to take a sample. Calculate a statistic, and use that statistic as some barometer of the population parameter. We're going to infer it to a population. But it is, somehow has to represent the patient, the population parameter. So, in essence what we're trying to do here, is with confidence intervals is to put bounds, upper and lower bounds, around the sample statistic that we've found. And we would hope that the true population parameter falls within those bounds. So let's look at an example from the literature itself. Dinh and colleagues, they looked at Maternal HIV Seroconversion during Pregnancy. Unfortunate situation. Now they had a sample of over 9,000, 9,802 mother-child pairs and they established a seroconversion rate. During pregnancy are about 3.3%. That was their mean. And they put it around that, on the 95% confidence intervals of 2.8 to 3.8%. What does that really mean? What does it mean that 2.8 to 3.8%? Which you would notice is on either side of the 3.3% that they found in this study. Now remember, we don't know what it really is out there in the large population. We do not examine the whole population, we don't investigate the whole population, it's impossible. We only have the data at hand from that sample. Now, it's a large sample, 9,802. But it really does refer only to the data that's at hand. That's the only mean we can really calculate. Now, we must infer that 3.3% to the larger population so that when a healthcare worker that's working on the same type of population, can use those results. So for instance, they have to come up with some sort of intervention program. They've got to use those results, they're got to know what the likelihood is, what the likely values are that the patient, population parameter might be. Now, how sure are we that the value that was found, the 3.3 was found, how sure are we that that really is the true patient, the population parameter out there. Well, if we chose the sample properly, if it was proper random selection and that sample does reflect the population, then that 3.3% must really be somewhere in the ball park. It must be close. And for that, we need this concept of confidence levels and we'll discuss that next.