Roadmap
Rearview Mirror
- Statisticians create a population model to represent the world
- A population model has parameters we are interested in
- Ex: A parameter might represent the effect that a vitamin has on test performance
- A null hypothesis is a specific statement about a parameter
- Ex: The vitamin has zero effect on performance
- A hypothesis test is a procedure for rejecting or not rejecting a null, such the probability of a type 1 error is constrained.
Today
- There are often multiple hypothesis tests you can apply to a scenario.
- Our primary concern is choosing a test with assumptions we can defend.
- Secondarily, we want to maximize power.
Looking ahead
- Next week, we start working with models for linear regression
- We will see how hypothesis testing is also used for regression parameters.