DOX Exam2020 Comprehensive

Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

Experimental Design: Comprehensive Exam (2020)

Name: Student id#: Department:

Please show your work in your answer sheet.

1. (20 points) An experiment was performed to compare the effectiveness of two diets
(Diet A and Diet B) as a strategy for losing weight. Sixteen subjects were recruited
for the study. Eight subjects were assigned to Diet A and the other eight were
assigned for Diet B in a completely randomized fashion. The weight loss in pounds
is reported by the subjects after three weeks shown below.

Weight loss (pounds) mean variance


Diet A 6, 6, 5, 2, 2, 3, 2, 0 ȳ1 = 3.25 S12 = 33.5/7 = 4.79
Diet B 4, 2, 2, 2, 1, -1, -2, -2 ȳ2 = 0.75 S22 = 33.5/7 = 4.79

Place your answers by hand.

(a) Write out the ANOVA table for the analysis of this completely randomized
experiment, and fill in the correct numbers of SS, df, MS and F0 . Does there
appear to be an effect of diet?
(b) Suppose it is now discovered that the groups were highly unbalanced with
respect to sex. Group Diet A had six males (6, 6, 5, 2, 2, 3) and two females
(2, 0), whereas group Diet B had two males (4, 2) and six females (2, 2, 1, -1,
-2, -2). Ignoring treatment, the results for males and females are shown below:
Weight loss (pounds) mean variance
Male 6, 6, 5, 2, 2, 3, 4, 2 ȳ1 = 3.75 S12 = 21.5/7 = 3.07
Female 2, 0, 2, 2, 1, -1, -2, -2 ȳ2 = 0.25 S22 = 21.5/7 = 3.07
Write out the ANOVA table to test for an effect of sex, ignoring the Diets.
Does there appear to be a sex effect?
(c) Two students were asked to investigate the combined effects of treatment and
sex using two-way ANOVA. The first student reported that the Diet effect is
significant:

df SS MS F0 p-value
Diet 1 25.000 25.000 7.5 0.01798
Sex 1 27.000 27.000 8.1 0.01473
Diet*Sex 1 0.000 0.000 0.0 1.00000
Residuals 12 40.000 3.333

1
The second student reported that the Diet effect is not significant:

df SS MS F0 p-value
Sex 1 49.000 49.000 14.7 0.002378
Diet 1 3.000 3.000 0.9 1.361497
Diet*Sex 1 0.000 0.000 0.0 1.000000
Residuals 12 40.000 3.333
Explain why the two answers are different. Which answer is better?

2. (20 points) An experiment is conducted to compare 4 navigation techniques (Factor


A, Fixed), 2 input methods (Factor B, Fixed), in 36 subjects (Factor C, Random).
Each subject is measured in each combination of levels of A and B once. Consider
the following unrestricted model:

yijk = µ + αi + βj + γk + (αβ)ij + (αγ)ik + (βγ)jk + ijk ,


Pa Pb Pa Pb iid
where i=1 αi = j=1 βj = i=1 (αβ)ij = j=1 (αβ)ij = 0, γk ∼ N (0, σγ2 ),
iid 2 iid 2 iid 2
(αγ)ik ∼ N (0, σαγ ), (βγ)jk ∼ N (0, σβγ ), ijk ∼ N (0, σ ), and all random
effects and errors are pairwise independent.

(a) Using the EMS rules, find the following E(M SA ), E(M SB ), E(M SC ), E(M SAB ),
E(M SAC ), E(M SBC ), and E(M SE ).
(b) Complete the following ANOVA table, testing all main effects and 2-factor
interactions.
Source df SS MS F0 dfnum dfden Fdfnum ,dfden
A 66996
B 30636
C 84008
AB 18710
AC 148797
BC 68605
Error 97282
Total 515034

3. (20 points) Consider the following SAS output from analysis of a balanced incom-
plete block design (BIBD). The statistical model is

yij = µ + τi + βj + ij .

We assume that all factors are fixed. τi and βj are respectively treatment and block
effects.

2
(a) Write out the remaining conditions for the above model.
(b) What are the hypotheses of interest? Should the hypothesis be rejected? Why,
or why not?
(c) If the grand mean is 72.50, compute τ̂1 , · · · , τ̂4 (treatment effect).

Source DF Sum of Squares Mean Square F Value Pr>F


Model 6 77.7500000 12.95833333 19.94 0.0024
Error 5 3.2500000 0.65000000
Total 11 81.0000000

Source DF Type III SS Mean Square F Value Pr>F


Block 3 66.08333333 22.02777778 33.89 .0010
Trt 3 22.75000000 7.58333333 11.67 .0107

Trt y LSMEAN Standard Error LSMEAN Number


1 71.37500000 0.4868051 1
2 71.62500000 0.4868051 2
3 72.00000000 0.4868051 3
4 75.00000000 0.4868051 4

4. (14 points) Among a population of lakes, the mean adult fish lengths are normally
distributed, with approximately 95% of the lake means lying between 50.2 and 69.8
centimeters. Within lakes, approximately 95% of the fish have lengths within 11.76
centimeters of the lake mean. Consider a one-way random effects model, where a
sample of g lakes is selected and n fish are sampled from each lake.

Yij = µ + τi + ij , τi ∼iid N (0, στ2 ), ij ∼iid N (0, σ 2 ),

for i = 1, . . . , g; j = 1, . . . , n, τi ’s and ij are independent.

(a) Obtain µ, στ2 , and σ 2 .


1
Pg Pn
(b) Give the expectation and variance of Ȳ·· = gn i=1 j=1 Yij .

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy