You are on page 1of 3

homework chapter 21

xiang liu
April 1st, 2015

Contents
Based on KNNL problem 20.2-20.4 on page 889. Coin-operated terminals. A university
computer service conducted an experiment in which one coin-operated computer terminal
was placed at each of four different locations on the campus last semester during the midterm
week and again during the final week of classes. The data that follow show the number of
hours each terminal was NOT in use during the week at the four locations (factor A) and for
the two different weeks (factor B).

Factor B
Factor A j=1 j=2
(location) Midterm Final
i=1 16.5 21.4
i=2 11.8 17.3
i=3 12.3 16.9
i=4 16.6 21.0

Assume that no-interaction ANOVA model (20.1) is appropriate.

(a) Plot the dat in the format of Figure 20.1. Does it appear that interaction effects are
present? Does it appear that factor A and factor B main effects are present?
of Hours not in Use

Location
20

1
Average Number

4
18

2
3
16
14
12

1 2

Week

1
No clear interaction effect presented since the four line seem to be parallel to each other.
Changing location wouldnt affect the relationship between week and average time of
no in use.
It does appear to have main effects for factor A and B respectively. Lower average
number of hours the terminal isnt in use is in week 1 (vs week 2) and location 2 & 3
(vs location 1 & 4)

(b) Set up the ANOVA table.

Table 1: Analysis of Variance Table

Df Sum Sq Mean Sq F value Pr(>F)


location 3 37.01 12.34 107.3 0.0015
week 1 47.05 47.05 409.1 3e-04
Residuals 3 0.345 0.115 NA NA

(c) Obtain the fitted value for cell i = 3 and j = 2.

32 = Y3. + Y.2 Y.. = 17.025

(d) Construct a 95% confidence interval for 32 .

Estimated Value Standard Error t 2.5% 97.5%


17.02 0.27 3.18 16.17 17.88

(d) Carry out tests for location and week effects; use a level of significance = 0.05. State
your conclusions.

Table 3: Analysis of Variance Table

Df Sum Sq Mean Sq F value Pr(>F)


location 3 37.01 12.34 107.3 0.0015
week 1 47.05 47.05 409.1 3e-04
Residuals 3 0.345 0.115 NA NA

2
Model: yij = + i + j + ij where 4i=1 i = 0 and 2j=1 j = 0 and ij iid(0, 2 )

Main effect for location

H0: all s equal 0 vs Ha: at least one i 6= 0

Test Statistics: as shown in ANOVA output in part (b), the F = M SMlocation


SE
= 107.3
and p-value < 0.05. Rejecting null hypothesis
Conclusion: We have enough evidence to prove that location affects average hour
of time when machine is not in use.

Main effect for week

H0: all s equal 0 vs Ha: at least one j 6= 0

Test Statistics: as shown in ANOVA output in part (b), the F = MMSSEweek


= 409.1
and p-value < 0.001. Rejecting null hypothesis
Conclusion: We have enough evidence to prove that average hour of time when
machine is not in use is different from week to week (week 1 vs week 2).

(e) Conduct the Tukeys test for additivity to decide if interaction effects can be ignored at
a level of significance = 0.05. State your conclusions.

Table 4: Tukeys one df F test for Additivity: location


and week on time

Test statistic num df denom df P value


0.5897 1 2 0.5228

H0:D = 0 (no interaction) vs Ha:D 6= 0 (there is interaction)


Since the test statistics F(1,2) = 0.59, p-value > 0.5. We cannot reject the null
hypothesis at = 0.05 significance level. The no-interaction model is plausible.
There is no interaction effect of location and week on average hour of time when machine
is not in use.

Based on KNNL problem 22.13 -22.14 on pages 943-944. A manufacturer of felt-tip


markers investigated by an experiment whether a proposed new display, featuring s picture
of a physician, is more effective in drugstores than the present counter display, featuring a
picture of an athlete and designed to be located in the stationery area. Fifteen drugstores
of similar characteristics were chosen for the study. They were assigned at random in equal
numbers to one of the following three treatments: (1) present counter display in stationery

You might also like