o If there are 2 groups, you’re deciding between a t-test or ANOVE
o If there are 3 groups, you would use ANOVA Independent variable or factor o The variable that produces your groups (ex: diagnostic group, treatment group, running program) Dependent variable or criterion o The variable that you are measuring as your outcome in each group Basic Design o Null hypothesis would be indicating that all means are equal Meaning that the population means will not differ from each other o Alternate hypothesis would indicate that there is at least one difference between the means Assumptions o Homogeneity of variance Each group has the same variance on the dependent variable in their population o Normality in each population Dependent variable would be continuous The kurtosis and skew don’t matter Interval or ratio data When deciding between ANOVA and t-test: o You would use ANOVA when you have independent variables with more than 2 groups If you don’t, you would have to do multiple t-tests When you do multiple t-tests, you’re also inflating the Type 1 error rate With an ANOVA test, the independent variable has more than 2 levels to compare and we are trying to see if they differ on a continuous dependent variable We would do a one-way ANOVA if we have 1 independent variable and one dependent variable Different Types of Variation: o Within-groups variation: variation due to chance An estimate of how much variation happens by chance Treatment or IV effect should be larger than within o Between-groups variation: variation due to chance and treatment (if any exists) Inherent variation aka error variation o Our estimate of chance variation 4 Steps of Hypothesis Testing o Step 1: Define null and research/alternative hypotheses o Step 2: Draw 2 or more random samples from the population and calculate observed F o Step 3:Describe the sampling distribution, given the null hypothesized population parameter and sample size o Step 4:Determine the cutoff value on the sampling distribution, compare our sample statistic to this cutoff, and decide whether to reject or retain the null hypothesis If we fail to reject/retain, we are saying that the differences between Group means are no different than we would expect by chance alone. Also saying there is no significant effect of the IV on the DV Also saying that the group means do not differ If we reject, group means are not all equal-there is a significant effect of the IV on the DV Also saying that group means are not all the same We can’t tell which group mean was different with the ANOVA test Remember: alpha is the type 1 error rate per test Way to cope with Type 1 error in post hoc tests o Limit the number of comparison o Adjust the alpha level We would use a Two-Way ANOVA when you have 2 independent variables o With a 2-way ANOVA test, you can look at the separate contributions of each independent variable and how they interact with each other Advantages of a 2-way ANOVA o Can determine if the 2 IVS interact