In: Statistics and Probability
In a collaborative trial, four laboratories were sent samples from the same batch of a pharmaceutical product and requested to perform ten assays and report the results based on percentage of a labeled amount of the drug
Lab | Score |
Lab A | 100 |
Lab A | 99.8 |
Lab A | 99.5 |
Lab A | 100.1 |
Lab A | 99.7 |
Lab A | 99.9 |
Lab A | 100.4 |
Lab A | 100 |
Lab A | 99.7 |
Lab A | 99.9 |
Lab B | 99.5 |
Lab B | 100 |
Lab B | 99.3 |
Lab B | 99.9 |
Lab B | 100.3 |
Lab B | 99.5 |
Lab B | 99.6 |
Lab B | 98.9 |
Lab B | 99.8 |
Lab B | 100.1 |
Lab C | 99.6 |
Lab C | 99.3 |
Lab C | 99.1 |
Lab C | 99.7 |
Lab C | 99.6 |
Lab C | 99.4 |
Lab C | 99.5 |
Lab C | 99.5 |
Lab C | 99.5 |
Lab C | 99.9 |
Lab D | 99.8 |
Lab D | 100.5 |
Lab D | 100 |
Lab D | 100.1 |
Lab D | 99.4 |
Lab D | 99.6 |
Lab D | 100.2 |
Lab D | 99.9 |
Lab D | 100.4 |
Lab D | 100.1 |
1-Which labs have means that are significantly different? (mark all that apply)
a. Lab A and Lab B |
||
b. Lab A and Lab C |
||
c. Lab A and Lab D |
||
d. Lab B and Lab C |
||
e. Lab B and Lab D |
||
f. Lab C and Lab D |
2- What is the conclusion of your hypothesis test?
a. All the means are different between the labs |
||
b. At least one of the labs has a mean that is significantly different from the others |
||
c. None of the above |
To answer the given questions, first we will have to conduct a one-way ANOVA and obtain the ANOVA table.
We have to test
H0: μA = μB = μC = μD
H1: at least one μi is different.
Original observation |
squared observation |
|||||||||
LAB A |
LAB B |
LAB C |
LAB D |
LAB A |
LAB B |
LAB C |
LAB D |
|||
100 |
99.5 |
99.6 |
99.8 |
10000 |
9900.25 |
9920.16 |
9960.04 |
|||
99.8 |
100 |
99.3 |
100.5 |
9960.04 |
10000 |
9860.49 |
10100.25 |
|||
99.5 |
99.3 |
99.1 |
100 |
9900.25 |
9860.49 |
9820.81 |
10000 |
|||
100.1 |
99.9 |
99.7 |
100.1 |
10020.01 |
9980.01 |
9940.09 |
10020.01 |
|||
99.7 |
100.3 |
99.6 |
99.4 |
9940.09 |
10060.09 |
9920.16 |
9880.36 |
|||
99.9 |
99.5 |
99.4 |
99.6 |
9980.01 |
9900.25 |
9880.36 |
9920.16 |
|||
100.4 |
99.6 |
99.5 |
100.2 |
10080.16 |
9920.16 |
9900.25 |
10040.04 |
|||
100 |
98.9 |
99.5 |
99.9 |
10000 |
9781.21 |
9900.25 |
9980.01 |
|||
99.7 |
99.8 |
99.5 |
100.4 |
9940.09 |
9960.04 |
9900.25 |
10080.16 |
|||
99.9 |
100.1 |
99.9 |
100.1 |
9980.01 |
10020.01 |
9980.01 |
10020.01 |
|||
Total |
999 |
996.9 |
995.1 |
1000 |
GT=3991 |
99800.66 |
99382.51 |
99022.83 |
100001 |
Σx2=398207.04 |
Mean |
99.9 |
99.69 |
99.51 |
100 |
We have n= 40
t=4
r=10
ANOVA Table
Source of Variation |
SS |
df |
MS |
F |
p value |
Between Groups |
1.437 |
3 |
0.48 |
4.82 |
0.006 |
Within groups (Error) |
3.578 |
36 |
0.10 |
||
Total |
5.015 |
39 |
In the above ANOVA table we have p-value= 0.006, which is less than 0.05 and it suggests that we have enough evidence against null hypothesis to reject the null and conclude that at least one of the labs has a mean that is significantly different from the others.
Answer(a): To know Which labs have means that are significantly different we need to do pairwise comparison of means using Tukey’s HSD as below:
The critical value for Tukey’s HSD can be obtained as below:
We have q=q(0.95,4,36)= 3.808798
MSE =0.1
r=10
Hence the critical value is
We have pairwise mean difference for all treatment groups which is presented below. If the magnitude of pairwise difference is greater than critical value, it is considered to be significantly different.
pair |
Pair wise difference of means |
S/NS |
Lab A - Lab B |
0.21 |
NS |
Lab A - Lab C |
0.39 |
S |
Lab A - Lab D |
0.1 |
NS |
Lab B - Lab C |
0.18 |
NS |
Lab B - Lab D |
0.31 |
NS |
Lab C - Lab D |
0.49 |
S |
We can see that the pair Lab A - Lab C and Lab C – Lab D are significantly different from each other while other pairs are not significantly different.
Hence the correct option is b and f
Answer(2):
We can obtain the conclusion of hypothesis test from the ANOVA table which we have already stated that at least one of the labs has a mean that is significantly different from the others.
Hence the correct option is b.