Questions:
Part One - Multiple Testing
Read Lecture Seven . The lectures from last week and Lecture Seven discuss issues around using a single test versus multiple uses of the same tests to answer questions about mean equality between groups. This suggests that we need to master-or at least understand-a number of statistical tests. Why can't we just master a single statistical test-such as the t-test-and use it in situations calling for mean equality decisions?
Part Two - ANOVA
Read Lecture Eight . Lecture Eight provides an ANOVA test showing that the mean salary for each job grade significantly differed. It then shows a technique to allow us to determine which pair or pairs of means actually differ. What other factors would you be interested in knowing if means differed by grade level? Why? Can you provide an ANOVA table showing these results? (Do not bother with which means differ.) How does this help answer our research question of equal pay for equal work? What kinds of results in your personal or professional lives could use the ANOVA test? Why?
Part Three - Effect Size
Read Lecture Nine . Lecture Nine introduces you to Effect size measure. There are two reasons we reject a null hypothesis. One is that the interaction of the variables causes significant differences to occur - our typical understanding of a rejected null hypothesis. The other is having a large sample size - virtually any difference can be made to appear significant if the sample is large enough. What is the Effect size measure? How does it help us decide what caused us reject the null hypothesis?