Estimating variance: should I use n or n - 1? The answer is not what you think

Estimates of population parameters based on samples are not exact: there is always some error involved. In principle, one can estimate a population parameter with any estimator, but some will be better than others. There is one particular case which was always very confusing to me (because of the multiple alternatives) and that is the estimation of the variance of a Normal population from a sample. In this article I describe the properties and tradeoffs among the different alternatives and discuss how important these differences are.

Maximum likelihood estimation from scratch

Maximum likelihood estimates of a distribution Maximum likelihood estimation (MLE) is a method to estimate the parameters of a random population given a sample. I described what this population means and its relationship to the sample in a previous post. Before we can look into MLE, we first need to understand the difference between probability and probability density for continuous variables. Probability density can be seen as a measure of relative probability, that is, values located in areas with higher probability will get have higher probability density.

Beyond normality: the bootstrap method for hypothesis testing

tl;dr: Parametric bootstrap methods can be used to test hypothesis and calculate p values while assuming any particular population distribution we may want. Non-parametric bootstrapping methods can be used to test hypotheses and calculate p values without having to assume any particular population as long as the sample can be assumed to be representative of the population and one can transform the data adequately to take into account the null hypothesis.

Where do p-values come from? Fundamental concepts and simulation approach

tl;dr: P-values are tail probabilities calculated from the sampling distribution of a sample-based statistic. This sampling distribution will depend on the size of the sample, the statistic being calculated and assumptions about the random population from which the data could have been sampled. For a few cases, analytical p-values are available and, for the rest of cases, approximations based on Monte Carlo simulation can be computed by generating the sampling distribution from the population.