Using a Different Criterion for Estimation of the Sample Mean

The sample mean, as we know it, is the maximum likelihood estimater of that best possible central location for the normal distribution. And what you wonderfully described in terms of derivatives, in setting em to zero and solving we talked about last week for the ordinary lease squares. We're doing a very similar kind of thing, but just with a different criterion. Instead of minimizing the sum of the squared residuals, we are going to maximize the likelihood that we would have observed our sample data exactly. It's just a different calvin ball rule.