Sie sind auf Seite 1von 2

Homework 4 (due Feb 16)

In your solution include your name and the homework number. Please staple your
pages together. When solving the problems below, give detailed derivations in order to
receive credit.

1. (Normal probabilities) Do Problem 10, p. 506 in Rice (2007).


2. (Normal probability plots) Use R to do the following simulations about normal prob-
ability plots. (Note the commands qqnorm, rnorm, rchisq, rt, runif, rlnorm.)
(a) Generate samples of sizes n = 20, n = 50 and n = 100 from a standard normal
distribution N (0, 1) and show the three normal probability plots. Repeat this
several times to get a feeling for how random variation affects normal probability
plots. You only need to turn in one set of plots from these several repeats.
(b) Repeat (a) for a χ2 -distribution with 2 degrees of freedom (i.e. an exponential
distribution).
(c) Repeat (a) for a χ2 -distribution with 6 degrees of freedom.
(d) Repeat (a) for a t-distribution with 2 degrees of freedom.
(e) Repeat (a) for a uniform distribution on [0, 1].
(f) Repeat (a) for log-normal distribution.
3. (Pairwise t-tests versus F-test) At http://www.stat.uchicago.edu/∼wbwu/24500/,
you find the text file hwk4-1.data.txt containing 4 columns, each of which contains
20 observations. Regard the i-th column as a sample from N (µi , σ 2 ), i = 1, . . . , 4.
Note that the population variance σ 2 is assumed to be the same for all four samples.
(a) Using R, perform all six two-sample t-tests for the hypotheses Hij : µi = µj ,
1 ≤ i < j ≤ 4. Which hypotheses would you reject at significance level α = 0.1?
(b) Now make a Bonferroni-adjustment to the significance level in order to control
the probability that at least one null hypothesis is falsely rejected at level 0.1.
Which of the tests in (a) would reject their null hypothesis if this Bonferroni-
adjusted significance level is employed?
(c) In order to test the hypothesis

H0 : µ1 = µ2 = µ3 = µ4

you consider the test statistic

T = max{|Tij | | 1 ≤ i < j ≤ 4},

where Tij is the test statistic from the two-sample t-test of the hypothesis Hij :
µi = µj tested in (a). The idea is to reject H0 if T is too large. Give a simple
upper bound on the p-value of the test based on T . Does this upper bound
allow you to reject H0 at level α = 0.1? In other words, is the upper bound
smaller than 0.1?

1
(d) Perform the F -test from the one-way analysis of variance to test the hypothesis
H0 considered in (c). Do you reject H0 at significance level α = 0.1?

4. (Data analysis: One-way ANOVA) Do Problem 21, P. 508, in Rice (2007). [You do
not need to do the nonparametric part]

5. Design a simulation study to compare the performance of kernel density estimates and
the density estimate based on parametric forms, under two cases (i) the parametric
form is correct and (ii) the parametric form is not correct.

Das könnte Ihnen auch gefallen