Professional Documents
Culture Documents
Dr. Stack
MATH 533 Graduate Student Portfolio Problems
Spring 2018
1. The nitrogen contents of red clover plants inoculated with five strains of Rhizobium are given in the
following table:
a. Is there evidence of a difference in the effects of the five treatments on the mean nitrogen content?
Analyze the data completely and draw conclusions based on your analysis. Use alpha = 0.01.
H0: µ1 = µ2 = µ3 = µ4= µ5
Ha: Not all the µi’s are equal
𝒅𝒇𝟏 = 𝒕 − 𝟏 = 𝟓 − 𝟏 = 𝟒
𝒅𝒇𝟐 = 𝒏𝒕 − 𝒕 = 𝟑𝟎 − 𝟓 = 𝟐𝟓 and 𝜶 =. 𝟎𝟏
CR: 𝑭 ≥ 𝟒. 𝟏𝟖
Based on our ANOVA test, our F statistic is ≈ 𝟐𝟑. 𝟒𝟒𝟐 with a p-value of ≈. 𝟎𝟎𝟎𝟏. Since our F-statistic lies in
the critical region we would reject Ho. Since our p-value is less than 𝜶 =. 𝟎𝟏, we would reject Ho. Therefore,
we can conclude that there is evidence of a difference in the effects of the five treatments on the mean
nitrogen content.
In order to see where the differences lie, we will use Fisher’s LSD because we do not have equal treatment
sample sizes:
𝟏 𝟏
𝑳𝑺𝑫 = 𝒕𝜶/𝟐 , 𝒗(√𝑴𝑺𝑬𝒓𝒓𝒐𝒓)(√𝒏 + 𝒏 )
𝒊 𝒋
.𝟎𝟏
𝒅𝒇 = 𝒏𝒕 − 𝒕 = 𝟑𝟎 − 𝟓 = 𝟐𝟓 with 𝜶 = =. 𝟎𝟎𝟓
𝟐
Hattie Burford
Dr. Stack
MATH 533 Graduate Student Portfolio Problems
Spring 2018
𝟏 𝟏
𝑳𝑺𝑫 𝒘𝒉𝒆𝒏 𝒏𝒊 𝒂𝒏𝒅 𝒏𝒋 𝒂𝒓𝒆 𝟒 𝒂𝒏𝒅 𝟔 = (𝟐. 𝟕𝟖𝟕)(√𝟑. 𝟏𝟓𝟔𝟗𝟒𝟏𝟒) (√ + ) ≈ 𝟑. 𝟏𝟗𝟔𝟒
𝟒 𝟔
𝟏 𝟏
𝑳𝑺𝑫 𝒘𝒉𝒆𝒏 𝒏𝒊 𝒂𝒏𝒅 𝒏𝒋 𝒂𝒓𝒆 𝟒 𝒂𝒏𝒅 𝟕 = (𝟐. 𝟕𝟖𝟕)(√𝟑. 𝟏𝟓𝟔𝟗𝟒𝟏𝟒) (√ + ) ≈ 𝟑. 𝟏𝟎𝟑𝟖
𝟒 𝟕
𝟏 𝟏
𝑳𝑺𝑫 𝒘𝒉𝒆𝒏 𝒏𝒊 𝒂𝒏𝒅 𝒏𝒋 𝒂𝒓𝒆 𝟒 𝒂𝒏𝒅 𝟖 = (𝟐. 𝟕𝟖𝟕)(√𝟑. 𝟏𝟓𝟔𝟗𝟒𝟏𝟒) (√ + ) ≈ 𝟑. 𝟎𝟑𝟐𝟒
𝟒 𝟖
𝟏 𝟏
𝑳𝑺𝑫 𝒘𝒉𝒆𝒏 𝒏𝒊 𝒂𝒏𝒅 𝒏𝒋 𝒂𝒓𝒆 𝟓 𝒂𝒏𝒅 𝟔 = (𝟐. 𝟕𝟖𝟕)(√𝟑. 𝟏𝟓𝟔𝟗𝟒𝟏𝟒) (√ + ) ≈ 𝟐. 𝟗𝟗𝟖𝟓
𝟓 𝟔
𝟏 𝟏
𝑳𝑺𝑫 𝒘𝒉𝒆𝒏 𝒏𝒊 𝒂𝒏𝒅 𝒏𝒋 𝒂𝒓𝒆 𝟓 𝒂𝒏𝒅 𝟕 = (𝟐. 𝟕𝟖𝟕)(√𝟑. 𝟏𝟓𝟔𝟗𝟒𝟏𝟒) (√ + ) ≈ 𝟐. 𝟖𝟗𝟗𝟓
𝟓 𝟕
𝟏 𝟏
𝑳𝑺𝑫 𝒘𝒉𝒆𝒏 𝒏𝒊 𝒂𝒏𝒅 𝒏𝒋 𝒂𝒓𝒆 𝟓 𝒂𝒏𝒅 𝟖 = (𝟐. 𝟕𝟖𝟕)(√𝟑. 𝟏𝟓𝟔𝟗𝟒𝟏𝟒) (√ + ) ≈ 𝟐. 𝟖𝟐𝟑𝟎
𝟓 𝟖
𝟏 𝟏
𝑳𝑺𝑫 𝒘𝒉𝒆𝒏 𝒏𝒊 𝒂𝒏𝒅 𝒏𝒋 𝒂𝒓𝒆 𝟔 𝒂𝒏𝒅 𝟕 = (𝟐. 𝟕𝟖𝟕)(√𝟑. 𝟏𝟓𝟔𝟗𝟒𝟏𝟒) (√ + ) ≈ 𝟐. 𝟕𝟓𝟓𝟎
𝟔 𝟕
𝟏 𝟏
𝑳𝑺𝑫 𝒘𝒉𝒆𝒏 𝒏𝒊 𝒂𝒏𝒅 𝒏𝒋 𝒂𝒓𝒆 𝟔 𝒂𝒏𝒅 𝟖 = (𝟐. 𝟕𝟖𝟕)(√𝟑. 𝟏𝟓𝟔𝟗𝟒𝟏𝟒) (√ + ) ≈ 𝟐. 𝟔𝟕𝟒𝟑
𝟔 𝟖
𝟏 𝟏
𝑳𝑺𝑫 𝒘𝒉𝒆𝒏 𝒏𝒊 𝒂𝒏𝒅 𝒏𝒋 𝒂𝒓𝒆 𝟕 𝒂𝒏𝒅 𝟖 = (𝟐. 𝟕𝟖𝟕)(√𝟑. 𝟏𝟓𝟔𝟗𝟒𝟏𝟒) (√ + ) ≈ 𝟐. 𝟓𝟔𝟐𝟖
𝟕 𝟖
We will now calculate the difference in sample means for each of the pairwise comparisons:
𝝁𝟏 𝒂𝒏𝒅 𝝁𝟐 :
̅̅̅̅̅̅̅̅̅
|𝒀̅𝒊 − ̅̅̅
𝒀𝒋 |=|𝟑𝟎. 𝟐𝟕𝟓 − 𝟐𝟏. 𝟔𝟖| = |𝟖. 𝟓𝟗𝟓| = 𝟖. 𝟓𝟗𝟓
𝟖. 𝟓𝟗𝟓 > 𝟑. 𝟑𝟐𝟏𝟖
Therefore, there is a significant difference between 𝝁𝟏 & 𝝁𝟐
𝝁𝟏 𝒂𝒏𝒅 𝝁𝟑 :
̅̅̅̅̅̅̅̅̅
|𝒀̅𝒊 − ̅̅̅
𝒀𝒋 |=|𝟑𝟎. 𝟐𝟕𝟓 − 𝟐𝟎. 𝟏𝟓| = |𝟏𝟎. 𝟏𝟐𝟓| = 𝟏𝟎. 𝟏𝟐𝟓
Hattie Burford
Dr. Stack
MATH 533 Graduate Student Portfolio Problems
Spring 2018
𝝁𝟏 𝒂𝒏𝒅 𝝁𝟒 :
̅̅̅̅̅̅̅̅̅
|𝒀̅𝒊 − ̅̅̅
𝒀𝒋 |=|𝟑𝟎. 𝟐𝟕𝟓 − 𝟐𝟏. 𝟐𝟕𝟏𝟒𝟐𝟗| = |𝟗. 𝟎𝟎𝟑𝟓𝟕𝟏| = 𝟗. 𝟎𝟎𝟑𝟓𝟕𝟏
𝟗. 𝟎𝟎𝟑𝟓𝟕𝟏 > 𝟑. 𝟏𝟎𝟑𝟖
Therefore, there is a significant difference between 𝝁𝟏 & 𝝁𝟒
𝝁𝟏 𝒂𝒏𝒅 𝝁𝟓 :
̅̅̅̅̅̅̅̅̅
|𝒀̅𝒊 − ̅̅̅
𝒀𝒋 |=|𝟑𝟎. 𝟐𝟕𝟓 − 𝟐𝟑. 𝟖𝟔𝟐𝟓| = |𝟔. 𝟒𝟏𝟐𝟓| = 𝟔. 𝟒𝟏𝟐𝟓
𝟔. 𝟒𝟏𝟐𝟓 > 𝟑. 𝟎𝟑𝟐𝟒
Therefore, there is a significant difference between 𝝁𝟏 & 𝝁𝟓
𝝁𝟐 𝒂𝒏𝒅 𝝁𝟑 :
̅̅̅̅̅̅̅̅̅
|𝒀̅𝒊 − ̅̅̅
𝒀𝒋 |=|𝟐𝟏. 𝟔𝟖 − 𝟐𝟎. 𝟏𝟓| = |𝟏. 𝟓𝟑| = 𝟏. 𝟓𝟑
𝟏. 𝟓𝟑 < 𝟐. 𝟗𝟗𝟖𝟓
Therefore, there is not a significant difference between 𝝁𝟐 & 𝝁𝟑
𝝁𝟐 𝒂𝒏𝒅 𝝁𝟒 :
|𝒀̅̅̅̅̅̅̅̅̅
̅𝒊 − ̅̅̅
𝒀𝒋 |=|𝟐𝟏. 𝟔𝟖 − 𝟐𝟏. 𝟐𝟕𝟏𝟒𝟐𝟗| = |. 𝟒𝟎𝟖𝟓𝟕𝟏| =. 𝟒𝟎𝟖𝟓𝟕𝟏
. 𝟒𝟎𝟖𝟓𝟕𝟏 < 𝟐. 𝟖𝟗𝟗𝟓
Therefore, there is not a significant difference between 𝝁𝟐 & 𝝁𝟒
𝝁𝟐 𝒂𝒏𝒅 𝝁𝟓 :
̅̅̅̅̅̅̅̅̅
|𝒀̅𝒊 − ̅̅̅
𝒀𝒋 |=|𝟐𝟏. 𝟔𝟖 − 𝟐𝟑. 𝟖𝟔𝟐𝟓| = |−𝟐. 𝟏𝟖𝟐𝟓| = 𝟐. 𝟏𝟖𝟐𝟓
𝟐. 𝟏𝟖𝟐𝟓 < 𝟐. 𝟖𝟐𝟑𝟎
Therefore, there is not a significant difference between 𝝁𝟐 & 𝝁𝟓
𝝁𝟑 𝒂𝒏𝒅 𝝁𝟒 :
̅̅̅̅̅̅̅̅̅
|𝒀̅𝒊 − ̅̅̅
𝒀𝒋 |=|𝟐𝟎. 𝟏𝟓 − 𝟐𝟏. 𝟐𝟕𝟏𝟒𝟐𝟗| = |−𝟏. 𝟏𝟐𝟏𝟒𝟐𝟗| = 𝟏. 𝟏𝟐𝟏𝟒𝟐𝟗
𝟏. 𝟏𝟐𝟏𝟒𝟐𝟗 < 𝟐. 𝟕𝟓𝟓𝟎
Therefore, there is not a significant difference between 𝝁𝟑 & 𝝁𝟒
𝝁𝟑 𝒂𝒏𝒅 𝝁𝟓 :
̅̅̅̅̅̅̅̅̅
|𝒀̅𝒊 − ̅̅̅
𝒀𝒋 |=|𝟐𝟎. 𝟏𝟓 − 𝟐𝟑. 𝟖𝟔𝟐𝟓| = |−𝟑. 𝟕𝟏𝟐𝟓| = 𝟑. 𝟕𝟏𝟐𝟓
𝟑. 𝟕𝟏𝟐𝟓 > 𝟐. 𝟔𝟕𝟒𝟑
Therefore, there is a significant difference between 𝝁𝟑 & 𝝁𝟓
𝝁𝟒 𝒂𝒏𝒅 𝝁𝟓 :
̅̅̅̅̅̅̅̅̅
|𝒀̅𝒊 − ̅̅̅
𝒀𝒋 |=|𝟐𝟏. 𝟐𝟕𝟏𝟒𝟐𝟗 − 𝟐𝟑. 𝟖𝟔𝟐𝟓| = |−𝟐. 𝟓𝟗𝟏𝟎𝟕𝟏| = 𝟐. 𝟓𝟗𝟏𝟎𝟕𝟏
𝟐. 𝟓𝟗𝟏𝟎𝟕𝟏 > 𝟐. 𝟓𝟔𝟐𝟖
Therefore, there is a significant difference between 𝝁𝟒 & 𝝁𝟓
Hattie Burford
Dr. Stack
MATH 533 Graduate Student Portfolio Problems
Spring 2018
In order to test if there was a violation in the required conditions to conduct the ANOVA, Hartley’s F
test was conducted.
Therefore, since our F statistic falls in the CR and our p-value is less than alpha, we would reject Ho,
meaning the conditions were not met.
Therefore, we may want to consider using a nonparametric test, such as the Kruskel-Wallis Test to test if:
Ho: the t distributions are identical
Ha: not all t distributions are identical
Using Stat Crunch to run the Kruskel-Wallis test:
df=4, Chi-squared=19.347235, and our p-value=.0007
Using table 7, our CR:
𝑿𝟐 ≥ 𝟏𝟑. 𝟐𝟖.
Therefore, we would reject Ho.
Hattie Burford
Dr. Stack
MATH 533 Graduate Student Portfolio Problems
Spring 2018
2. Biofeedback is a treatment technique in which people are trained to improve their health by using
signals from their own bodies. Specialists in many different fields use biofeedback to help their
patients cope with pain. A study was conducted to compare a biofeedback treatment for chronic pain
with an NSAID medical treatment. A group of 2,000 newly diagnosed chronic pain patients were
randomly assigned to receive one of the two treatments. After six weeks of treatments, the pain level
of the patients was assessed with the following results:
b. Is there significant evidence (alpha = 0.05) of a difference in the two treatments relative to the
proportions of patients who experienced a significant reduction in pain?
Using my 2-prop Z test on my TI83:
x1=780, x2=690
n1,2=1000
𝒑𝟏 ≠ 𝒑𝟐
Z’=4.55996
p=.000005
̂𝟏=.78
𝒑
̂𝟐=.69
𝒑
̂=.735
𝒑
Hattie Burford
Dr. Stack
MATH 533 Graduate Student Portfolio Problems
Spring 2018
Therefore, since our p-value is approximately 0, less than alpha=.05, we reject Ho. Our proportions for the
two treatments are statistically different.
3. A petroleum company was interested in comparing the miles per gallon achieved by four different
gasoline blends (A, B, C, and D). Because there can be considerable variability due to differences in
driving characteristics and car models, these two extraneous sources of variability were included as
“blocking” variables in the study. The researcher selected four different brands of cars and four
different drivers. The drivers and brands of cars were assigned to blends in the manner displayed in
the following table. The mileage (in mpg) obtained over each test run was recorded as follows:
Car Model
Driver 1 2 3 4
1 A(14.5) B(32.8) C(12.7) D(24.2)
2 B(15.3) C(26.4) D(15.1) A(25.5)
3 C(13.5) D(33.5) A(19.5) B(32.1)
4 D(16.0) A(31.5) B(15.7) C(27.6)
a. Write a model for this experimental setting.
Ho: 𝝁𝑨 = 𝝁𝑩 = 𝝁𝑪 = 𝝁𝑫
Ha: At least one of the means is not equal
We are testing the means to determine if the choice in gasoline blend affects the miles per gallon
achieved.
Source Df SS MS F’ P-Value
Rows 3 40.8319 13.6106 3.8862 .074003
Columns 3 803.3219 267.77397 76.4566 .000036
Treatment 3 32.2719 10.7573 3.0715 .1124
Error 6 21.0137 3.5023
Total 15 897.4394
c. What conclusions can you draw concerning the best gasoline blend?
Since our p-value (for treatments) is not less than alpha=.05, we would FAIL TO REJECT. This means
that we conclude that, statistically, our means of our treatments are equal.
d. Compute the relative efficiency of the Latin square design relative to a completely randomized design.
Interpret this value.
𝑴𝑺𝑹 + 𝑴𝑺𝑪 + (𝒕 − 𝟏)𝑴𝑺𝑬
𝑹𝑬 =
(𝒕 + 𝟏)(𝑴𝑺𝑬)
Therefore, we would need about 17 x more data in using a completely randomized design to have the same
efficiency as this Latin square.
e. If future studies were to be conducted, would you recommend using both car model and driver as
blocking variables? Explain.
Because we found our relative efficiency to be rather large, we can assume that using the model and driver
as blocking variables would be beneficial in future studies, since we found this design to be efficient.
4. An experiment was set up to compare the effect of different soil pH and calcium additives on the
increase in trunk diameters for orange trees. Annual applications of elemental sulfur, gypsum, soda
ash, and other ingredients were applied to provide pH value levels of 4, 5, 6, and 7. Three levels of a
calcium supplement (100, 200, and 300 pounds per acre) were also applied. All factor-level
combinations of these two variables were used in the experiment. At the end of a 2-year period, four
diameters were examined at each factor-level combination. The data appear in the table below:
Calcium
pH Value 100 200 300
4.0 5.2 7.4 6.3
5.9 7.0 6.7
6.3 7.6 6.1
6.1 7.2 6.4
5.0 7.1 7.4 7.3
Hattie Burford
Dr. Stack
MATH 533 Graduate Student Portfolio Problems
Spring 2018
c. Perform an analysis of variance (create the ANOVA table) and identify the experimental design. Let
alpha = 0.05 and run the test.
Using the Two-Way ANOVA on Stat Crunch (AOV table for the complete factorial design):
𝑭′ = 𝟏𝟏. 𝟑𝟓𝟔𝟔𝟏𝟖
𝒑 − 𝒗𝒂𝒍𝒖𝒆 <. 𝟎𝟎𝟎𝟏