Unformatted text preview: STAT 4220
HW7 (20 points)
Due on Thu, April 1
1. To compare the eﬀects of ﬁve diﬀerent assembly methods (denoted by the
Latin letters A, B , C , D, and E ) on the throughput, an experiment based
on a GraecoLatin square was conducted which involved three blocking
variables: day, operator, and machine type. The data are given in Table 1,
where the machine type is denoted by the ﬁve Greek letters. The response,
throughput, is the number of completed pieces per day and is given in the
parentheses in the table. Day
1
2
3
4
5 Aα
Bγ
C
Dβ
Eδ 1
(102)
(92)
(96)
(120)
(123) Table 1: Throughput Data
Operator
2
3
4
Bβ (105) Cγ (82) Dδ (141)
Cδ (112) D
(131) Eα (112)
Dα (130) Eβ (108) Aγ (73)
Eγ (100) Aδ (111) B
(116)
A
(110) Bα (111) Cβ (85) 5
E
Aβ
Bδ
Cα
Dγ (132)
(99)
(129)
(100)
(100) (a) (4+3 points) Analyze the data and compare the ﬁve assembly methods. Give the ANOVA Table. Do multiple comparisons to ﬁnd which
of the methods are signiﬁcantly better than the others?
(b) (3 points) Which factors in the ANOVA table are not signiﬁcant?
Can you ﬁt a better model without the insigniﬁcant factors? Explain
clearly.
2. Natrella (1963, pp. 13–14) described an experiment on a resistor mounted
on a ceramic plate in which the impact of four geometrical shapes of the
resistors on the current noise of the resistor is studied. Only three resistors can be mounted on one plate. The design and data for the resistor
experiment are given in Table 2.
Table 2: Data, Resistor Experiment
Shape
Plate
A
B
C
D
1
1.11
0.95 0.82
2
1.70 1.22
0.97
3
1.60 1.11 1.52
4
1.22 1.54 1.18
(a) (3 points) Describe and justify the design.
(b) (7 points) Analyze the experiment. 1 ...
View
Full
Document
This note was uploaded on 06/06/2011 for the course STAT 4220 taught by Professor Smith during the Spring '08 term at UGA.
 Spring '08
 smith

Click to edit the document details