STAT 509-E01 Spring 2017 HW 5 Sol

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

HW 5-1 (Uploaded on Apr.

20, 2017) Name:

HW 5.1. Inferences of µ1 − µ2 with known σ1 and σ2 . The concentration of active ingredient in


a liquid laundry detergent is thought to be affected by the type of catalyst used in the process. The
concentration is known to have normal distribution with the standard deviation 3 grams per liter
regardless of the catalyst type. Then observations on concentration are taken with each catalyst,
and the data follow:
Catalyst 1: 57.9, 66.2, 65.4, 65.4, 65.2, 62.6, 67.6, 63.7, 67.2, 71.0.
Catalyst 2: 66.4, 71.7, 70.3, 69.3, 64.8, 69.6, 68.6, 69.4, 65.3, 68.8.

(a) Find a 95% confidence interval on the difference in mean active concentrations for the two
catalysts.

(b) With significant level α = 0.05 , is there any evidence to indicate that the mean active con-
centrations depend on the choice of catalyst (i.e., they are different or not)? What is your
conclusion?

Sol

(a) Since population variances are known, we use 2-sampZInt in calculator or using the equation
at page 100 in lecture notes to obtain the confidence interval. A 95% confidence interval for the
difference in mean active concentrations for the two catalysts is (−5.83, −0.5704).
Interpretation: We are 95% confident that the population mean difference of concentrations
µ1 − µ2 is in the interval (−5.83, −0.5704).

(b) Null and alternative hypotheses:

H0 : µ1 = µ2 versus Ha : µ1 6= µ2 .

which are equivalent to

H0 : µ1 − µ2 = 0 versus Ha : µ1 − µ2 6= 0.

Set significant level α = 0.05.

• Confidence Interval Approach: According to the reject criterion on page 101, since “0” is
not in 95% confidence interval, we do reject the null hypothesis.
• P-value approach: Here P-value can be obtaining by 2-SampleZTest (since population
variances are known) in calculator. Since P-value= 0.0170 < 0.05 = α, we do reject the
null hypothesis.

Conclusion: At significant level α = 0.05, the data do provide sufficient evidence to conclude
that the real difference of concentrations µ1 − µ2 is different from 0, i.e., µ1 is different from µ2 .

1
HW 5.2. Inferences of µ1 − µ2 with σ1 = σ2 . An article in Nature (2003, Vol. 48, p. 1013)
described an experiment in which subjects consumed different types of chocolate to determine the
effect of eating chocolate on a measure of cardiovascular health. We will consider the results for
only dark chocolate and milk chocolate. In the experiment, 12 subjects consumed 100 grams of dark
chocolate and 200 grams of milk chocolate, one type of chocolate per day, and after one hour, the
total antioxidant capacity of their blood plasma was measures in an assay. The subjects consisted of
seven women and five men with an average age range of 32.2±1 years, an average weight of 65.8±3.1
kg, and average body mass index of 21.9 ± 0.4 kh/m2 . Data similar to that reported in the article
follows.
Dark Chocolate: 118.8, 122.6, 115.6, 113.6, 119.5, 115.9, 115.8, 115.1, 116.9, 115.4, 115.6, 107.9.
Milk Chocolate: 102.1, 105.8, 99.6, 102.7, 98.8, 100.9, 102.8, 98.7, 94.7, 97.8, 99.7, 98.6.
Let µ1 be the mean blood plasma antioxidant capacity resulting from eating dark chocolate and µ2
be the mean blood plasma antioxidant capacity resulting from eating milk chocolate.

(a) According to the box plots in Figure 1 or sample standard deviations, do you believe that the
population standard deviations are equal? Why?

(b) Calculate a two-sided 95% on the mean difference µ1 −µ2 of blood plasma antioxidant capacities.

(c) Is there evidence to support the claim that consuming dark chocolate produces a higher mean
level of total blood plasma antioxidant capacity that consuming milk chocolate? Use significant
level α = 0.05. What is your conclusion?

(d) Do normal qq plots of blood plasma antioxidant capacity in Figure 2 indicate any violations of
the assumptions for the tests and confidence interval that your performed?

Sol.

(a) Yes, since the quartile differences of two samples look similarly so we treat σ1 = σ2 .

(b) Since population variances are unknown but equal, we use 2-sampTInt in calculator with Pooled
option or using the equation at page 103 in lecture notes to obtain the confidence interval. A
95% confidence interval for the difference in mean active concentrations for the two catalysts is
(13.142, 18.608).
Interpretation: We are 95% confident that the mean difference µ1 − µ2 of blood plasma antioxi-
dant capacities is in (13.142, 18.608).

(c) Null and alternative hypotheses:

H0 : µ1 = µ2 versus Ha : µ1 > µ2 .

which is equivalent to
H0 : µ1 − µ2 = 0 versus Ha : µ1 − µ2 > 0.

Set significant level α = 0.05.

2
• Confidence Interval Approach: According to the reject criterion on page 104 in lecture
notes, we construct the confidence lower-bound which is 13.612 (or we can use 2-sampTInt
by setting confidence level to be 2α = 0.1 and without Pooled option, then the smaller
number of the output confidence is the 95%lower-bound). Since δ0 = 0 < 13.612 which
does not meet the reject criterion, we do reject the null hypothesis.
• P-value approach: Here P-value can be obtaining by 2-SampTTest with Pooled option (since
population variances are unknown but equal) in calculator. Since P-value= 1.8417×10−11 <
0.05 = α, we do reject the null hypothesis.

Conclusion: At significant level α = 0.05, the data do provide sufficient evidence to conclude
that the real difference of concentrations µ1 − µ2 is larger than 0, i.e., µ1 is larger than µ2 .

(d) The first sample may not come from normal distribution since there are points departuring the
straight line. In the second qq plot, all points are close to the straight so the normal assumption
for the second sample is appropriate.

3
115
105
95

Dark Milk

Figure 1: Boxplots for Chocolate data.

Normal Q−Q Plot Normal Q−Q Plot


106
120
Sample Quantiles

Sample Quantiles
102
115

96 98
110

−1.5 −0.5 0.5 1.5 −1.5 −0.5 0.5 1.5


Theoretical Quantiles Theoretical Quantiles

Figure 2: Normal quantile-quantile (qq) plots for Chocolate data.

4
HW 5.3. Inferences of µ1 − µ2 with σ1 6= σ2 . An article in Polymer Degradation and Stability
(2006, Vol. 91) presented data from a nine-year aging study on S537 foam. Foam samples were
compressed to 50% of their original thickness and stored at different temperatures for nine years.
At the star of the experiment as well as during each year, sample thickness was measured, and the
thicknesses of the eight samples at each storage condition were recorded. The data of two storage
conditions follow:
50◦ C: 0.047, 0.060, 0.061, 0.064, 0.080, 0.090, 0.118, 0.165, 0.183.
60◦ C: 0.062, 0.105, 0.118, 0.137, 0.153, 0.197, 0.210, 0.250, 0.375.

(a) According to the box plots in Figure 3 or standard deviations, what do you believe: σ1 = σ2 or
σ1 6= σ2 ? Why?

(b) Find a 95% confidence interval for the difference in the mean compression for the two tempera-
tures.

(c) Is there evidence to support the claim that mean compression increases with the temperature at
the storage condition?

(d) Do normal qq plots of compression in Figure 4 indicate any violations of the assumptions for the
tests and confidence interval that your performed?

Sol.

(a) Since the quartile differences of two samples look differently, we treat σ1 6= σ2 .

(b) Since population variances are unknown and unequal, we use 2-sampTtest in calculator without
Pooled option or using the equation at page 106 in lecture notes to obtain the confidence interval.
A 95% confidence interval for the difference in mean active concentrations for the two catalysts
is (−0.1587, −0.0055).
We are 95% confident that the difference of population means µ1 − µ2 of thicknesses is in
(−0.1587, −0.0055).

(c) Null and alternative hypotheses:

H0 : µ1 = µ2 versus Ha : µ1 < µ2 ,

which are equivalent to

H0 : µ1 − µ2 = 0 versus Ha : µ1 − µ2 < 0.

Set significant level α = 0.05.

• Confidence Interval Approach: According to the reject criterion on page 107 in lecture notes,
we construct the confidence upper-bound which is −0.0194 (or we can use 2-sampTInt by
setting confidence level to be 2α = 0.1 and without Pooled option, then the larger number

5
of the output confidence is the 95% upper-bound). Since δ0 > −0.0194 which does meet
the rejection criterion, we do reject the null hypothesis.
• P-value approach: Here P-value can be obtaining by 2-SampTTest without Pooled option
(since population variances are unknown and not equal) in calculator. Since P-value=
0.0189 < 0.05 = α, we do reject the null hypothesis.

Conclusion: At significant level α = 0.05, the data do provide sufficient evidence to conclude
that the population mean difference of thicknesses µ1 − µ2 is smaller than 0, i.e., µ1 is smaller
than µ2 .

(d) The first qq plot looks fine since all points are not too far away from the straight line. In the
second qq plot, the departure on the right-hand side is acceptable. So the normal assumptions
for two samples are appropriate.

6
0.35
0.25
0.15
0.05

50 60

Figure 3: Boxplots for Foam data.

Normal Q−Q Plot Normal Q−Q Plot


0.18

0.35
Sample Quantiles

Sample Quantiles
0.14

0.25
0.10

0.15
0.06

0.05

−1.5 −0.5 0.5 1.5 −1.5 −0.5 0.5 1.5


Theoretical Quantiles Theoretical Quantiles

Figure 4: Normal quantile-quantile (qq) plots for Foam data.

7
HW 5.4. Inference of p1 −p2 . An article in Knee Surgery, sports traumatology, Arthroscopy (2005,
Vol. 13, pp. 273-279) considered arthroscopic meniscal repair with an absorbable screw. Results
showed that for tears greater than 25 millimeters, 14 of 18 (78%) repairs were successful, but for
shorter tears, 22 of 30 (73%) repairs were successful.

(a) Calculate a one-sided 95% confidence bound on the difference in proportions.

(b) Is there evidence that the success rate is greater for longer tears? Use significant level α = 0.05.
What is your conclusion?

Sol.

(a) We use TwoSamp-PropInt in calculator or confidence interval stated on page 110 in lecture to
obtain the confidence interval. A 95% confidence interval for the the population proportion
difference p1 − p2 of success rates is (−0.2044, 0.293).
Interpretation: We are 95% confident that the population proportion difference p1 − p2 of success
rates is in (−0.2044, 0.293).

(b) Null and alternative hypotheses:

H0 : p1 − p2 = 0 versus Ha : p1 − p2 > 0,

which are equivalent to


H0 : p1 = p2 versus Ha : p1 > p2 ,

Set significant level α = 0.05.

• P-value approach: Here P-value can be obtaining by 2-PropZTest in calculator. Since


P-value= 0.3653 > 0.05 = α, we do not reject the null hypothesis.

Conclusion: At significant level α = 0.05, the data do not provide sufficient evidence to conclude
that the population proportions difference of success rate p1 − p2 is larger than 0, i.e., p1 is not
larger than p2 .

8
HW 5.5. Inferences of σ12 /σ22 . The thickness of a plastic film (in mils) on a substrate material
is thought to be influenced by the temperature at which the coating ins applied. In completely
randomized experiment, 11 substracteds are coated at 125◦ F, resulting in a sample mean coating
thickness of sample mean x1 = 103.5 and a sample standard deviation of s1 = 10.2 Another 13
substrates are coated at 150◦ F for which sample mean x2 = 99.7 and sample standard deviation
s2 = 20.1 are observed. It was originally suspected that raising the process temperature would
reduce mean coating thickness.

(a) Construct a 95% confidence interval for σ12 /σ22 .

(b) Test H0 : σ1 = σ2 against Ha : σ1 > σ2 using α = 0.05. What is your conclusion?

Sol.

(a) Here we have α = 0.05, s21 = 10.22 , and s22 = 20.12 , a 95% confidence interval for σ12 /σ22 is given
by

s21 s21
 
1
× , × Fn2 −1,n1 −1,α/2
s22 Fn1 −1,n2 −1,α/2 s22
(10.2)2 (10.2)2
 
1
= × , × 3.6209
(20.1)2 3.3736 (20.1)2
= (0.0763, 0.9325)

Interpretation: We are 95% confident that the ratio σ12 /σ22 is in (0.0763, 0.9325).

(b) Null and alternative hypotheses:

H0 : σ1 = σ2 versus Ha : σ1 6= σ2 .

Set significant level α = 0.05.

• Confidence Interval Approach: According to the reject criterion on page 120, since “1” is
not in the 95% confidence interval (0.0763, 0.9325), we do reject the null hypothesis.
• P-value approach: Here P-value can be obtaining by 2-SampFTest in calculator. Since
P-value= 0.0395 > 0.05 = α, we do not reject the null hypothesis.

Conclusion: At significant level α = 0.05, the data do not provide sufficient evidence that the
standard deviations are different.

9
HW 5.6. Inferences of µ1 − µ2 in matched pairs design. The manager of a fleet of automobiles
is testing two brands of radial tires and assigns one tire of each brand at random to the two rear
wheels of eight cars and runs the cars until the tires wear out. The data (in kilometers) follow:

(a) Find a 95% confidence interval on the difference in mean life.

(b) Which brand would you prefer? Test H0 : µD = 0 against Ha : µD 6= 0 using α = 0.05. What is
your conclusion?

Sol. First we calculate the difference, Brand 1 - Brand 2, for each car:

2607, 3020, 740, 150, −805, 560, 390, 285.

According to those differences, we construct one sample confidence interval for µD with unknown
population variance and perform one sample t−test

(a) We use T-Interval in calculator or confidence interval stated on page 122 in lecture to obtain the
confidence interval. We are 95% confident that the population mean difference µD of distances
until tires wearing out is in (−210.1197, 1946.8697).

(b) Null and alternative hypotheses:

H0 : µD = 0 versus Ha : µD > 0.

Set significant level α = 0.05.

• Confidence Approach:According to the reject criterion on page 123 in lecture notes, we


construct the confidence upper-bound which is ?? (or we can use T-Interval by setting
confidence level to be 2α = 0.1 and without Pooled option, then the larger number of the
output confidence is the 95% upper-bound).
• P-value approach: Here P-value can be obtaining by T-test in calculator. Since P-value=
0.0986 > 0.05 = α, we do not reject the null hypothesis.

Conclusion: At significant level α = 0.05, the data do not provide sufficient evidence to conclude
that the population mean difference µD of distances until tires wearing out is different.

10
Even though data do not provide sufficient evidence to tell there are difference between brands,
I prefer Brain 1 since most of the differences,283 Brain 1 - Brain 2, are larger than 0.

11
HW 5.7. The compressive strength of concrete is being studied, and four different mixing tech-
niques are being investigated. The following data have been collected. Test the hypothesis that
mixing techniques affect the strength of the concrete. Use α = 0.05.

Sol. Null and alternative hypotheses:

H0 : µ1 = µ2 = µ3 = µ4 .
versus
Ha : Not all population means are equal.

Set significant level α = 0.05.

• P-value approach: Here P-value can be obtaining by ANOVA in calculator. It can also be obtain
by the ANOVA table provided by R:

Analysis of Variance Table

Response: Compressive Strength


Df Sum Sq Mean Sq F value Pr(>F)
Mixing Technique 3 489740 163247 12.728 0.0004887 ***
Residuals 12 153908 12826
---
Signif. codes:
0 *** 0.001 ** 0.01 * 0.05 . 0.1 1

Since P-value= 4.887 × 10−4 < 0.05 = α, we do reject the null hypothesis.

Conclusion: At significant level α = 0.05, the data do provide sufficient evidence to conclude that
the population means of strength of the concrete are different due to mixing technique.

12
HW 5.8. An electronics engineer is interested in the effect on tube conductivity of five different
types of coating for cathode ray tubes in a telecommunications system display device. The following
conductivity data are obtained. Is there any difference in conductivity due to coating type? Use
α = 0.01.

Sol. Null and alternative hypotheses:

H0 : µ1 = µ2 = µ3 = µ4 .
versus
Ha : Not all population means are equal.

Set significant level α = 0.01.

• P-value approach: Here P-value can be obtaining by ANOVA in calculator. It can also be obtain
by the ANOVA table provided by R:

Analysis of Variance Table

Response: Conductivity
Df Sum Sq Mean Sq F value Pr(>F)
Coating Type 4 1060.50 265.125 16.349 2.414e-05 ***
Residuals 15 243.25 16.217
---
Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1

Since P-value= 2.41 × 10−5 < 0.01 = α, we do reject the null hypothesis.

Conclusion: At significant level α = 0.01, the data do provide sufficient evidence to conclude that
the population means of conductivity are different due to coating type.

13

You might also like