Null Hypothesis

This is the hypothesis that your treatment does nothing at all. (Refer to pp 5-6 of the Intro to Statistical Hypothesis Testing notes).

In hypothesis testing, one of the hypotheses to test is that nothing happens.

If you have a procedure to raise IQ scores and your treatment group has a mean IQ score of 100, there are 3 possible outcomes. 1, your treatment raises the scores, 2, your treatment lowers the scores, or 3, your treatment does nothing (the null hypothesis).

Knowing that, you can use the CLT to test the null hypothesis. This results in what's on page 7.

You can use this as you collect data. If you compare your actual observed mean to the sample distribution and look at how rare your observed mean is compared to the null hypothesis. If the observed mean would be very rare based on the null hypothesis, then the null hypothesis may be wrong - meaning that my treatment worked.

Decision rules let me decide how confident I want to be.