POLI 30 Lecture Notes - Lecture 16: Null Hypothesis, Observational Error, Sampling Error
Document Summary
Firslecture sixteen: introduction to inference and hypothesis testing 03/04/19. We use hypothesis testing to evaluate hypotheses derived from our theories. We acknowledge that there is random sampling error and that data could have a misleading trend. We rely on probability theory to answer the following fundamental question: For us this means that our theory is assumed wrong until we prove it. Second convict if guilty beyond a reasonable doubt. For us this means that the evidence-data- has to be consistent with our theory and inconsistent with the idea that our theory is wrong. Set alpha level - decide what reasonable doubt means. Set ahead of time- what probability is so low that you will reject the null hypothesis? alpha. There are situations where you want bigger, and where you want smaller. There is no relationship between x and y, or iv and dv or whatever we observe is just luck . Our theory is correct- there is a relationship.