Reply to thread

Message: [QUOTE="MadDogRevival, post: 2753931, member: 8028"] Let's suppose that you think that lilac pollen is causing bee die off. You begin your analysis by saying: "Lilac pollen is causing bee die off." That is your null hypothesis. You then seek to show that, statistically, lilac poison is NOT causing bee die off. You might review data on a thousand bee hives, only 20% of which have access to lilac poison. If, statistically, there is no (significant) correlation between hives dying off and lilac access, you have DISPROVED the null hypothesis. For example, if the numbers show that bee die off rate is the same for hives having access to lilac and hives that don't have access, then there would be no statistical connection shown in the data you have. Ergo, you have disproved or weakened the null hypothesis. Note that a null hypothesis can never by proved. In other words, if you, if fact, found a statistically significant connection between bee die off and access to lilac, you wouldn't (and couldn't) say that you "proved" the null hypothesis - no. What you would say is that null hypothesis could not be rejected (disproved). A null hypothesis is generally taken as "true" when you begin examining data, and, as you progress, you either reject it or you fail to reject it, depending on what the data say. In our case, my null hypothesis is: "Kevin Ollie is a terrible coach." If he goes on to do a good job coaching somewhere else, I'll reject the null hypothesis - something else must explain the last 3 years at UConn. If he goes no to do a lousy job coaching somewhere else, then I will not reject the null hypothesis and it will stand, pending further data/experimentation. Summary for casual fans: Frank thinks KO is a lousy coach and double dares him to prove it by not sucking at his next stop. Hat tip to CHief. [/QUOTE]

Verification: First name of men's bb coach