March 2019
Beginner to intermediate
448 pages
13h 14m
English
In the following exercise, we will generate two samples with the same mean and different variances, and we will calculate the effective error I rates for the t-test. In general, this is a futile exercise, since statistical tests are designed to yield the very same error I rate as they promise. However, in this case, we are introducing a violation to the assumptions in the test, namely that the variances are different (so the effective
rates will differ from the nominal ones).
If the variances were the same, we would expect to see that 5% of them have a p-value less than 0.05 ((in this case, even though the null hypothesis is ...
Read now
Unlock full access