Reference no: EM132232346
Question 1. For a two-way contingency table, let Πij be the cell probability for the (i, j)th cell for i=1,...,I; j=1,...,J.
a) For n= Σij nij independent observations, write down the multinomial probability that n11 fall in the (1, 1 )th cell,..., nij fall in the (i, j )th cell, ..., nIJ fall in the (I, J )th cell.
b) Let eij be the expected counts for the cell (i,j) under H0 hypothesis. Show that the likelihood- ratio test for two way contingency table is given by
G2 = 2 ∑ij nijln(nij/eij) = -2 ∑ij nijln(eij/nij).
c) Based on Taylor's expansion of natural log at x= 1 below
Ln(x) ≈ (x-1) - (x-1)2/2,
show that likelihood-ratio test can be approximated by
∑ij (nij - eij)2/nij .
[Hint: Apply Taylor expansion of Ln(x) at x = eij/nij in the second expression of G2.]
d) Let X be row random variable with categories indexed by i=1,..,I. Let Y be column random variable with categories indexed by j=1,..,J. Show that the degree of freedom of the likelihood- ratio test for independence between X and Y is (I-1)(J-1).
Question 2. The conditional distribution of Y given X for a 2 by 2 table is given below
Show that X and Y are independent if and only if Π1 = Π2.
Question 3. Suppose we have 2 random categorical variables X and Y whose underlying probability structure for the joint distribution is given by
|
Y
|
1
|
2
|
X
|
1
|
Π11
|
Π12
|
2
|
Π21
|
Π22
|
and we are interested in testing the hypothesis: H0 : Π11 = η (1-η), Π12 = (1-η)2, Π21= η2, Π22 = η (1- η) where 0 < η < 1. Do the following:
a) Find the marginal distributions of X and Y under H0 in terms of η.
b) Find the conditional distribution of Y at X = x under H0 in terms of η for x=1,2, respectively.
c) Show that X and Y are independent under H0.
d) Suppose we have following data nij's from a multinomial sampling
Find the MLE of η under H0.
e) For the data in (d), find the (estimated) expected value of nij for cell (i,j) under H0 for i=1,2;,j=1,2.
f) For the data in (d), conduct the Pearson χ2 for testing H0 and test H0 at level α = 0.05 using this test statistic.
g) For the data in (d), conduct the likelihood ratio test for testing H0 and test H0 at level α = 0.05 using this test statistic.
Question 4. In the United States, the estimated annual probability that a woman over the age of 35 dies of lung cancer equals 0.001304 for current smokers and 0:000121 for nonsmokers (M.Pagano and K. Gauvreau, Principles of Biostatistics, 1993, p. 134).
(a) Calculate and interpret the difference of proportions and the relative risk. Which of these measures is more informative for these data? Why?
(b) Calculate and interpret the odds ratio. Explain why the relative risk and odds ratio take similar values.
Question 5. The following table was taken from the 1991 General Social Survey.
|
Belief in Afterlife
|
Race |
Yes
|
No or Undecided
|
White |
621
|
239
|
Black |
89
|
42
|
a) Identify each classification as a response or explanatory variable.
b) Calculate sample i) odds ratio ii) relative risk iii) difference in proportions. Interpret the direction and strength of association based on odds ratio.
c) Obtain a 95% confidence interval for i) odds ratio, ii) relative risk iii) difference in proportions. Test the independence between belief in afterlife and race.
Note that
A large-sample confidence interval for the log of the relative risk is
log(p1/p2)±zα/2√((1-p1)/(n1p1) + (1-p2)/(n2p2))
where n1 = n1+ and n2 = n2+; and p1 = n11/n1+ and p2 = n21/n2+.
d) Are the conclusions in i), ii) and iii) consistent?
Question 6. In an article about crime in the United States, Newsweek magazine (Jan. 10, 1994) quoted FBI statistics stating that of all blacks slain in 1992, 94% were slain by blacks, and of all whites slain in 1992, 83% were slain by whites. Let Y denote the race of victim and X denote race of murderer.
(a) Which conditional distribution do these statistics refer to, Y given X, or X given Y ?
(b) Calculate and interpret the odds ratio between X and Y.
(c) Given that a murderer was white, can you estimate the probability that the victim was white? What additional information would you need to do this?