Hello, I'm a bit confused by the Examiner's Report answer for Question 6 Part I in the Sept 2013 paper. They have decided that the test statistic should be: (2nd Sample variance) / (1st Sample variance) Although everything I've come across so far does it the other way around. Am I missing something with regards to this particular question which means that I should work with the 2nd over the first? This is particularly annoying because one way will lead you to accept the Null hypothesis and the other way will lead to rejection. Thanks in advance for looking at this.
The ratio can be either way. You can take (S2/S1) or (S1/S2). But you have to opt for that value of the F-tabulated which is in line with the ratio you have taken. If you opt for (S1/S2) the F-tabulated will have df (n1-1,n2-1) and if you opt for (S2/S1) the F-tabulated will have df (n2-1,n1-1). Refer to this section in chapter 9. You will get a clear idea of why this is true. Hope this clears the doubt. If not, do repost!
