Hypothesis Testing Using a Null Model Many students have difficulty with understanding the logic behind hypothesis testing. I’ve found that requiring students to draw a clearly labeled null model while performing tests clarifies the entire process. A null model is simply a model of the sampling distribution if the null hypothesis were true. General steps: 1. State the null and alt hyp – (condish) 2. Draw a Null Model 3. Check your sample with the model. Could it have happened by random sampling variability? Example one: In 2000, 18% of ninth graders smoked cigarettes, has this changed? You collect a sample of 38 students and find that only 5 smoke….. What can you conclude? Suppose that in your sample 2 smoked… Would that be significant? This is just a general introduction to Null Models. If used well, they really impact a students conceptual understanding of hypothesis testing. Null models can be used for all of the tests that we run (Z and T).