ABSTRACT
Statistical testing has allowed the reader to discriminate perceived differences due to random variation from true differences due to a condition. But true differences supported by a statistical test do not necessarily imply that the difference is meaningful. The main assumption of the tests the people learned is that the sample measurements are normally distributed. Social media success is (partly) based on the concept of confirmation bias. An honest statistician should perform something called multiple testing correction. Random effect tests are a bit more complex as the people have to deal with two levels of uncertainty, that of the independent and that of the dependent variable. Mixed effects models are one of the most active areas of research in statistical analysis. No one is expected to do all statistical tests by hand. Thankfully, there are a number of statistical computing programs that facilitate the task.
