Let x1, x2, x3,..., xn be a random sample from a beta distribution with parameter .
Suppose that = = . To decide between two simple hypotheses
H0:
=1,
H1:
=2
One way to decide between H0 and H1 is to compare the corresponding likelihood functions:
l0=L(x1,x2,...,xn ; 0 ) ,
l1=L(x1,x2,...,xn ; 1 ).
More specifically, if l0 is much larger than l1, we should accept H0. On the other hand if l1 is much larger, we tend to reject H0. Therefore, we can look at the ratio of l0 and l1 to decide between H0 and H1. This is the idea behind likelihood ratio tests
we define ratio
= l0/l1 = = /
To perform a likelihood ratio test (LRT), we choose a constant c. We reject H0 if <c and accept it if c. The value of c can be chosen based on the desired .
We have given Beta distribution with = =
= l0 / l1
For H0: =1 thus l0 = = =
l0 = 1
and H1: =2 thus l1 = =
=
Now = 6
Thus l1 = 6 * x * (1 - x )
Hence = l0 / l1 = 1 / 6 * x * (1 - x )
Now we accept H0 if c
= 1 / 6 * x * (1 - x ) c
6 * x * (1 - x ) 1 / c
x - x2 1 / (6c)
where c is the threshold
Thus, we accept H0 if
x * (1 - x ) 1 / (6c)
Let us define c'=1 / (6c), where c' is a new threshold.
Remember that x is the observed value of the random variable x.
Thus, we can summarize the decision rule as follows. We accept H0 if
x * (1 - x )≤c'.
To choose c' we use the required Prob ( Type I error ) = alp ( given level of significance )
P(type I error) = P( Reject H0 | H0 : =1 ) = P( [x * (1 - x )] > c' | H0 ) = alp
Question 1: (10 marks) Let Y, Y....,Y, be a random sample from the beta distribution with...
Question: Let Y, Y be a random sample of size n=2 from a distribution with Pdf f(y; 6) = (6)e-0 OLGLD and o elsewhere We reject Ho : O=2 and accept H1:0=1 if the observed values of Y la are such that: fu, ; 2) fűz; 2) at f(4,;1) fŲz;1) 1 * Find the Significance level and the Power of the test when Ho is falso; given that {0:0=1,2}.
2. (20pts) Let Xi,..., X be a random sample from a population with pdf f(x)--(1 , where θ > 0 and x > 1. (a) Carry out the likelihood ratio tests of Ho : θ-a, versus Hi : θ a-show that the likelihod ratio statistic corresponding to this test, A, can be re-written as Λ = cYne-ouY, where Y Σ:.. In (X), and the constant c depends on n and θο but not on Y. (b) Make a sketch of...
Question 4 15 marks] The random variables X1, ... , Xn random variables with common pdf independent and identically distributed are 0 E fx (x;01) 0 independent of the random variables Y^,..., Y, which and are indepen are dent and identically distributed random variables with common pdf 0 fy (y; 02) 0 (a) Show that the MLE8 of 01 and 02 are 1 = X i=1 Y (b) Show that the MLE of 0 when 01 = 0, = 0...
Likelihood Ratio Tests - I only require (a) and (b) here. I'll post (c) and (d) for another question Let X1,..., Xn be a random sample from the distribution with pdf { 0-1e--)e f(r μ, θ ) - 0. where E Rand 0 > 0 (a) If 0 is known but a is unknown, find a likelihood ratio test (LRT) of size a for testing Η : μ> Ho Ho Ho versus where oi a known constant (b) If 0...
Likelihood Ratio Tests - I only require (a) and (b) here. I'll post (c) and (d) for another question Let X1,..., Xn be a random sample from the distribution with pdf { 0-1e--)e f(r μ, θ ) - 0. where E Rand 0 > 0 (a) If 0 is known but a is unknown, find a likelihood ratio test (LRT) of size a for testing Η : μ> Ho Ho Ho versus where oi a known constant (b) If 0...
Likelihood Ratio Tests - I only require (c) and (d) here. I have posted (a) and (b) in another question Let X1,..., Xn be a random sample from the distribution with pdf { 0-1e--)e f(r μ, θ ) - 0. where E Rand 0 > 0 (a) If 0 is known but a is unknown, find a likelihood ratio test (LRT) of size a for testing Η : μ> Ho Ho Ho versus where oi a known constant (b) If...
3. Suppose that X (X...,X) is a random sample from a uniform distribution of the interval [0,0], where the value of ? is unknown, and it is desired to test the hypotheses H: 0>2 [5] (a) Show that the uniform family f(x;0)-(1/0)1 om(r) : ? > 0 maxi-isnXi. has a monotone likelihood ratio in the statistic T(X)- X. whereX (n) [5] (b) Find a uniformly most powerful (UMP) test of level ? for testing Ho versus HI
Let X1,X be a random sample from an EXP(0) distribution (0 > 0) You will use the following facts for this question: Fact 1: If X EXP(0) then 2X/0~x(2). Fact 2: If V V, are a random sample from a x2(k) distribution then V V (nk) (a) Suppose that we wish to test Ho : 0 against H : 0 = 0, where 01 is specified and 0, > Oo. Show that the likelihood ratio statistic AE, O0,0)f(E)/ f (x;0,)...
Consider a random sample X1, ..., Xn from a normal distribution with known mean 0 and unknown variance 0 = 02 (a) Write the likelihood and log-likelihood function (b) Derive the maximum likelihood estimator for 6 (c) Show that the Fisher information matrix is I(O) = 2014 (d) What is the variance of the maximum likelihood estimator for @? Does it attain the Cramer-Rao lower bound? (e) Suppose that you are testing 0 = 1 versus the alternative 0 #...
2. (a) Suppose that x1,... , Vn are a random sample from a gamma distribution with shape parameter α and rate parameter λ, Here α > 0 and λ > 0. Let θ-(α, β). Determine the log-likelihood, 00), and a 2-dimensional sufficient statistic for the data (b) Suppose that xi, ,Xn are a random sample from a U(-9,0) distribution. f(x; 8) otherwise Here θ > 0, Determine the likelihood, L(0), and a one-dimensional sufficient statistic. Note that the likelihood should...