assign1 - where 2 is the critical value for a 2...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
PSTAT 120C: Assignment # 1 Due April 9, 2009 These problems will be due in lecture next Thursday. 1. Suppose that we have 9 independent observations from a normal distribution with standard deviation 10. We wish to test H 0 : μ = 150 vs. H A : μ 6 = 150 The GLRT with level α = 0 . 05 has the critical set C 1 = { ¯ x < 143 . 5 or ¯ x > 156 . 5 } . (a) Calculate the power of this test against the alternative μ = 151. (b) Calculate the power of this test against the alternative μ = 123. (c) Alternatively, consider the test with C 2 = { ¯ x < 144 . 5 or ¯ x > 160 . 9 } . i. Calculate the level of this second test. ii. Calculate the power against the alternatives μ = 151 and μ = 123. (d) Which of the two tests do you think is better? Why? 2. This problem is not due until April 16 (Exercise 10.105 on page 554 in WMS .) Let Y 1 ,...,Y n denote a random sample from a normal distribution with mean μ (unknown) and variance σ 2 . For testing H 0 : σ 2 = σ 2 0 against H A : σ 2 > σ 2 0 , show that the likelihood ratio test is equivalent to the χ 2 test with critical region C = ± S 2 ( n - 1) σ 2 0 > χ 2
Background image of page 1
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: where 2 is the critical value for a 2 distribution with n-1 degrees of freedom. 3. Suppose that Y is a random sample of size 1 from a population with density f ( y | ) = ( y -1 , y 1 , elsewhere. where &gt; 0. (a) Sketch the power function of the test with the critical region Y &gt; . 5. (Compute the power at a reasonable number of points and draw a curve that seems to t those values. Please be neat.) (b) Find the best test for testing H : = 1 versus H A : = 2 of size . ( ? ) Show that you get the same test for any alternative where H A : = t and t &gt; 1. 4. Show that if T is a sucient statistic for estimating from the data X 1 ,...,X n then the Generalized Likelihood Ratio Test (GLRT) statistic for testing H : versus H a : 6 is a function of T . (Hint: use the factorization theorem.) 1...
View Full Document

Ask a homework question - tutors are online