PYB210 Lecture Notes - Lecture 10: Type I And Type Ii Errors
Document Summary
What anova does and does not tell us. Planned comparison (a priori comparisons: chapter 4 & 7. 5. Controlling type i error rates: chapter 6. Remember that a significant f tells us that at least one of the sample means differs from the other sample means but it does not tells us which of these particular means differ. Ideally we will be able to predict some kind of directional hypothesis before conducting our experiment in which we predict which means will differ. We test these specific directional predictions using planned comparisons (a priori comparisons or linear contrasts). In this case, there is no need to report the overall anova. If however you cannot predict which means will likely differ then you should perform an overall anova to see if the iv has an effect. You would then perform post hoc comparisons to see which means differ. Smaller, more simplified versions of an anova which typically compare just two levels.