I have a 2 x 2 experimental design and have bootstrapped the distribution of a particular statistic for each cell. Is it possible to perform something akin to an ANOVA on these bootstrapped data such that I can characterize the main effect of each factor and their interaction? My intuition is that simply entering these bootstrapped values into a conventional ANOVA would be incorrect.
Asked
Active
Viewed 3,006 times
1 Answers
2
You must perform an ANOVA on each of the bootstrapped datasets. Take the resulting distribution of test statistics and apply appropriate inference. See also: Computing p-value using bootstrap with R

AdamO
- 52,330
- 5
- 104
- 209