In: Statistics and Probability
Application
Exercise:
A team of researchers recruited a random sample of migraine
sufferers. Before the study began at baseline, participants were
asked to record the duration of headaches they experienced in a
week. A few days later, all participants were asked to practice a
week-long relaxation technique designed to reduce migraine
headaches. A two-week off period in which the participants
practiced nothing was given between the relaxation technique.
Throughout the relaxation technique participants were asked to
record the duration of headaches. What can be concluded with an
α of 0.05?
Baseline | Week 1 | Week 3 |
9 7 12 8 12 7 11 7 10 |
9 7 6 6 9 5 6 5 9 |
8 |
e) Regardless of the H0
decision in b), conduct Tukey's post hoc test for
the following comparisons:
1 vs. 3: difference =_______________ ; significant: ---Select---Yes
OR No
1 vs. 2: difference =_______________ ; significant: ---Select---Yes
OR No
f) Regardless of the H0
decision in b), conduct Scheffe's post hoc test
for the following comparisons:
1 vs. 3: test statistic = ______________ ; significant:
---Select---Yes OR No
2 vs. 3: test statistic = ______________ ; significant:
---Select---Yes OR No
We are given the following data and are asked to conduct Tukey's post hoc test and Scheffe's post hoc for the following comparisons
Baseline | Week 1 | Week 3 |
---|---|---|
9 | 9 | 8 |
7 | 7 | 9 |
12 | 6 | 6 |
8 | 6 | 4 |
12 | 9 | 7 |
7 | 5 | 4 |
11 | 6 | 6 |
7 | 5 | 5 |
10 | 9 | 8 |
e) Tukey's post hoc analysis for the following comparisons:
1 vs 3 and 1 vs 2
To conduct Tukey's post hoc analysis we have to conduct ANOVA test so that we can get MSwithin value and the q statistic for the Tukey's post hoc analysis
Baseline | Week 1 | Week 3 | Ti. | ||
---|---|---|---|---|---|
9 | 9 | 8 | 26 | 225.33 | 226 |
7 | 7 | 9 | 23 | 176.33 | 179 |
12 | 6 | 6 | 24 | 192 | 216 |
8 | 6 | 4 | 18 | 108 | 116 |
12 | 9 | 7 | 28 | 261.33 | 274 |
7 | 5 | 4 | 16 | 85.33 | 90 |
11 | 6 | 6 | 23 | 176.33 | 193 |
7 | 5 | 5 | 17 | 96.33 | 99 |
10 | 9 | 8 | 27 | 243 | 245 |
Totals |
Correction Factor
ANOVA Table
Source of Variation | S.S | d.f. | M.SS | F-ratio |
---|---|---|---|---|
Between Process(Treatments) | 52.72 | (k-1=3-1) 2 | ||
Within Process (error) | 74.02 | (N-k=27-3) 24 | ||
Total | 126.74 | (N-1) 26 |
Now that we have constructed the ANOVA table, we can proceed with Tukey's post hoc calculation.
For that we have to calculate absolute difference of means of 1 and 3, 1 and 2 groups.
Baseline | Week 1 | Week 3 |
---|---|---|
9 | 9 | 8 |
7 | 7 | 9 |
12 | 6 | 6 |
8 | 6 | 4 |
12 | 9 | 7 |
7 | 5 | 4 |
11 | 6 | 6 |
7 | 5 | 5 |
10 | 9 | 8 |
Sum=83 | Sum=62 | Sum=57 |
, ,
1 vs 3. difference=9.22-6.33=2.89
1 vs 2 difference=9.22-6.89=2.33
Here n is the group size not the overall size
Since 1 vs 3 difference is larger than HSD value, we say that the difference is significant
Since 1 vs 2 difference is larger than HSD value, we say that the difference is significant
Hence we say that 1 vs 3 difference is significant and 1 vs 2 difference is also significant
f) We are asked to conduct Scheffe's post hoc test for the following comparisons
1 vs 3 and 2 vs 3.
First we find the absolute mean difference between 1 vs and between 2 vs 3.
Since we have already calculated absolute mean difference of 1 v3, we only need to calculate absolute mean difference of 2 vs 3.
2 vs 3 difference=6.89-6.33=0.56
We use the following formula to get Scheffe formula values.
=
Since the sample size is the same for all the three groups, one calculation applies to all other comparisons.
Comparing Scheffe formula value with 1 vs 3: test statistic is not significant because 1 vs 3 difference(2.89) is less than Scheffe formula(3.42).
Comparing Scheefe formula value with 2 vs 3: test statistic is not significant because 2 vs 3 difference(0.56) is less than Scheffe formula(3.42)
Larger values are statistically significant at the chosen significant level(0.05 in our case).
Hence 1 vs 3: test statistic is not significant and 2 vs 3: test statistic is not significant.