I'm a novice in statistics and I have some confusion about the assumption of independence for statistical tests.
- I searched the Internet and some information says that for the t-test, the observations in the two groups should be independent (that is, measurements in sample 1 and measurements in sample 2 should be different). Some other information says that all observations (even in the same group) should be independent. Which one is correct?
- Is the independence assumption for ANOVA and the independence assumption for the t-test the same?
- Do non-parametric tests, such as the Wilcoxon signed rank test, also need to satisfy the independence assumption?