 Hey Amy, I need help Hi Joey What's up? I don't understand this fail to reject the null stuff Why can't I just say I accept the null and fail to reject is a double negative anyway? Actually, it is a triple negative Fail, reject and null are all negatives But the grammar issue is not as important as the logic of fail to reject Except the null seems logical to me. Why is it wrong? Remember, this is about a test of a hypothesis about a characteristic of a population We are interested in because we don't have all the data about the population We must make do with sample data From that we make inferences about the population characteristic. I get that part We calculate a test statistic from the sample data and compare that with the value for the population parameter We assume to be true and see I even remembered the rule You gave me that population is to parameter as sample is to statistic You sure did so if the sample does not provide enough evidence to reject the null Doesn't that prove the null is true? No, it doesn't Remember that probability is involved Even a well-designed sample might not have the exact same characteristics as the population Huh, so even if the sample statistic is not different enough from the assumed population parameter for us to reject the null hypothesis assumption That does not mean our assumption is true. Correct There is still a lot of data about the population we do not have So the null hypothesis could still be wrong if we had all the missing data about the population We could know for sure But without that missing data, we just do not know if the null is true Wait That reminds me of the black swan idea Yes At one time people thought that all swans were white based on seeing only white swans in europe for centuries But in 1697 a dutch explorer found a black swan in australia and brought back the feathers to prove it And with that single data point the swans are white hypothesis was proved to be wrong Okay, I get it. We say fail to reject the null because there might still be a black swan out there for our null hypothesis too That's right As long as we have to use samples There is a possibility of making an error in our decision about the null hypothesis That brings us to my next question How does that type 1 and type 2 error thing work? We'll talk about that in another video See you later gator. Okay, the stats files here i come Thanks for watching