ABSTRACT
Hypothesis testing is like digging for treasure on a treasure island. The significance level sets the probability that we have actually found treasure rather than made a mistake. We are very conservative here (that is why we only accept a 5 in 100 chance of making a mistake). We do not wander about picking up any old piece of rusting metal we chance upon and claim that we have found treasure. Our fellow treasure hunters would soon get fed up with us. We want to be sure that when we claim to have found treasure then we are correct. In hypothesis testing we do not want to make a Type I error: that is, claim that we have found a significant difference between the population distributions when there is not one. We do not want to claim that we have found treasure when we have not. That is why we set the significance level at a small probability level.
