This is mostly a vocabulary note. [[Statistics Index|statistics]] is annoying enough as is, it's doubly-annoying when it hides fairly straightforward concepts behind esoteric and non-obvious terminology. Great example of crap terminology and phrasing: if your data don't support the idea you're testing (or don't rule out that it may not be true) you say: > "fail to reject the null hypothesis " This is literally saying it's "not not not true." Anyway. Hypothesis testing is the process of drawing two contrasting propositions relating the value of a population parameter (e.g. mean, variances, proportion, etc), one of which is assumed to be true absent contradictory data. **Null Hypothesis** - what we believe to be true unless we're given contradictory data **Alternate Hypothesis** - what we are testing against, the opposite of the null hypothesis **One-sample testing** - testing for a parameter of a given population **Two-sample testing** - testing to **compare** two populations, you draw one sample from each **Type 1 Error** - Testing rejects the null hypothesis, but it is actually true **Type 2 Error** - Testing fails to reject the null hypothesis, but it it should have been rejected # Example Null Hypothesis (H0): Individuals spend at least 10 hours per week in their vehicles Alt Hypothesis (H1): Individuals spend less than 10 hours per week in their vehicles This uses the _lower-tailed_ t-test. **** ## Source - Grad School lecture materials ## Related