How is the null hypothesis chosen? Why is it null?
Suppose you want to design an experiment to test whether two variables are related in a certain way. What you actually do is calculate the probability, based on your data, that the variables are not related in the way you suspect. The hypothesis that the variables are not related is called the null hypothesis (null because there is no relation). For example suppose you think that people who listen to heavy metal music are smarter than the general population. Your null hypothesis might be that the mean of the distribution of IQ scores amongst heavy metal listeners is less than 100 (that is, heavy metal music and IQ are not related in the way you thought). You would then have to collect data of IQ scores from heavy metal listeners, and in some way analyze this data to determine the probability that it was sampled from a distribution with mean 100 (this is where the real math comes in). In most cases you will set a threshold before you collect the data for when you’ll say that indeed your
