In: Statistics and Probability
An experiment was done to compare three different methods to inoculate entozoic amoebae. The response variable was number amoebae yield per culture.
Method 1 | Method 2 | Method 3 |
224 | 187 | 215 |
257 | 225 | 192 |
259 | 247 | 211 |
273 | 202 | 225 |
272 | 220 | 198 |
246 | 238 | 194 |
255 | 239 | 182 |
205 | 204 | 222 |
255 | 179 | |
236 |
Can you conclude, at the 5% level that the mean yield per culture differs among the inoculation methods (yes or no)?
What is the test statistic?
What is the p-value?
Find a 95% confidence interval for the individual μᵢ for Method 1.
When would you conclude at the 5% level that the mean yield for Method 1 is different than the mean yield for Method 2?
A. When the p-value for the F-test of Methods is less than .05 B. When the confidence interval for the mean yield from Method 1 does not overlap the confidence interval for the mean yield of Method 2 C. When the p-value for the F-test of Methods is less than .05, and the confidence interval for the mean yield from Method 1 does not overlap the confidence interval for the mean yield of Method 2 D. When the p-value for the F-test of Methods is less than .05, and the sample mean yield for Method 1 is different than the sample Mean Yield for Method 2
|
If the 95% confidence Method 2 mean yield was (211.672, 238.928) and the 95% confidence interval for the Method 3 mean yield was (187.63, 216.365) answer the following TRUE/FALSE questions.
· TRUE or FALSE: Method 1 and Method 2 have significantly different mean yields. · TRUE or FALSE: Method 1 and Method 3 have significantly different mean yields. · TRUE or FALSE: Method 2 and Method 3 have significantly different mean yields. · TRUE or FALSE: None of the methods have significantly different mean yields. |
Solution
Let xij = number amoebae yield per culture of the jth sample under Method i.
Part (a)
To test if the mean yield per culture differs among the inoculation methods
The test employed is: ANOVA (Analysis of Variance) – one way with unequal number of observations per treatment.
Final answers are given below. Back-up Theory and Details of Calculations follow at the end.
Sub-part (i)
yes Answer 1
Sub-part (ii)
Test statistic = 10.6761 Answer 2
Sub-part (iii)
p-value = 0.000482 Answer 3
Sub-part (iv)
95% confidence interval for the individual μᵢ for Method 1 is: [205.765, 291.985] Answer 4
[CI = X1bar ± (t24, 0.025 x s), where
X1bar = mean yield for Method 1 = 1991/8 = 248.875
t24, 0.025 = upper 2.5% point of t-distribution with 24 degrees of freedom = 2.0639
s = standard error = sqrt(MSE of ANOVA) sqrt(436.2906) = 20.8876]
Sub-part (v)
One would conclude at the 5% level that the mean yield for Method 1 is different than the mean yield for Method
When the p-value for the F-test of Methods is less than .05, and the confidence interval for the mean yield from Method 1 does not overlap the confidence interval for the mean yield of Method 2. Option C Answer 5
Part (b)
If the 95% confidence Method 2 mean yield was (211.672, 238.928) and the 95% confidence interval for the Method 3 mean yield was (187.63, 216.365),
Method 1 and Method 2 have significantly different mean yields. FALSE Answer 6 [given CI’s concern Method 2 and Method 3] Method 1 and Method 3 have significantly different mean yields. FALSE: Answer 7 [given CI’s concern Method 2 and Method 3] Method 2 and Method 3 have significantly different mean yields. FALSE: Answer 8 [given CI’s overlap] None of the methods have significantly different mean yields. FALSE: Answer 9 [given CI’s concern only Method 2 and Method 3 and Method 1 is not considered] |
Back-up Theory and Details of Calculations for ANOVA
Suppose we have data of a 1-way classification ANOVA, with r rows, and n observations per cell.
Let xij represent the jth observation in the ith row, j = 1,2,…,n; i = 1,2,……,r
Then the ANOVA model is: xij = µ + αi + εij, where µ = common effect, αi = effect of ith row, and εij is the error component which is assumed to be Normally Distributed with mean 0 and variance σ2.
Hypotheses:
Null hypothesis: H0: α1 = α2 = ….. = αr = 0 Vs Alternative: H1: at least one αi is different from other αi’s.
Now, to work out the solution,
Terminology:
Row total = xi.= sum over j of xij
Grand total = G = sum over i of xi.
Correction Factor = C = G2/N, where N = total number of observations = r x n
Total Sum of Squares: SST = (sum over i,j of xij2) – C
Row Sum of Squares: SSR = {(sum over i of xi.2)/n} – C
Error Sum of Squares: SSE = SST – SSR
Mean Sum of Squares = Sum of squares/Degrees of Freedom
Degrees of Freedom:
Total: N (i.e., rn) – 1;
Rows: (r - 1);
Error: Total - Row
Fobs: MSR/MSE;
Fcrit: upper α% point of F-Distribution with degrees of freedom n1 and n2, where n1 is the DF for the numerator MS and n2 is the DF for the denominator MS of Fobs
Significance: Fobs is significant if Fobs > Fcrit
Calculations
alpha |
0.05 |
#treat |
3 |
n1 |
8 |
n2 |
10 |
n3 |
9 |
N |
27 |
x1. |
1991 |
x2. |
2253 |
x3. |
1818 |
G = x.. |
6062 |
C |
1361031.2593 |
Sx1j^2 |
499385 |
Sx2j^2 |
511909 |
Sx3j^2 |
369524 |
Sxij^2 |
1380818 |
Sxi.^2/ni |
1370347.0250 |
SST |
19786.74074 |
SSR |
9315.765741 |
SSE |
10470.975 |
ANOVA |
Table |
alpha |
0.05 |
|||
Source |
df |
SS |
MS |
F |
Fcrit |
p-value |
Shift |
2 |
9315.76574 |
4657.88287 |
10.6761 |
3.402826 |
0.0004823 |
Error |
24 |
10470.98 |
436.290625 |
|||
Total |
26 |
19786.7407 |
761.02849 |
Since F > Fcrit or equivalently, since p-value < alpha, the null hypothesis is rejected.
Hence, we conclude that
the mean yield per culture differs among the inoculation methods
DONE