Know for Quiz #3 on A&A chapters 8&9:
THIS WILL BE ALL ABOUT T-TESTS (nothing about ANOVA)
1. There are three kinds of T-tests discussed in the 2 chapters. Know the situation in which we use each one (single sample t, dependent means t, independent means t) If I describe a situation, you should be able to identify which test is appropriate (including a z-test, which is what we use when population variance IS known). Read the description of situations in your handouts.
2. Be able to describe the differences between a normal curve and a t-distribution. Know what happens as the df get larger, and be able to draw a comparison picture showing the differences between how normal curve and t (with small df) look.
3. Know WHAT is different in hypothesis testing for the different kinds of t-tests. You do NOT need to memorize formulas for messy things like pooled variance. But you SHOULD know which test uses pooled variance, which uses difference scores, which has two sample means versus one sample mean and a known population mean, and so on. Compare your handouts for the three tests and get clear on where steps are different.
4. Know why we use N-1 when estimating variance from the sample: We lose a degree of freedom once we calculate the sample mean, and this also corrects for the tendency of sample variance to underestimate population variance.
5. Know the assumptions for t-tests: A. We assume that the underlying population is normal. B. We assume the variance is equal between two populations (when there are two populations). When do we really have to worry about violating assumption A? When the underlying population is seriously skewed, especially if we are using a one-tailed test.