In: Statistics and Probability
The following data are from a completely randomized design.
|
a) For the CRD, we have various sum of squares defined as following
Sum of Squares, Treatment | |
Sum of Squares, Error | |
Mean Squares, Treatment | |
Mean Squares, Error |
where is the jth observation of ith treatment, is the ith treatment mean and is the grand mean.
k= number of treatments, n= number of observation of each treatment and k-1 is degree of freedom(d.f.) of treatment sum of square and k(n-1) is d.f. of error sum of square.
using the values given we can compute,
Sum of Squares, Treatment | 563.3 |
Sum of Squares, Error | 66 |
Mean Squares, Treatment | 281.7 |
Mean Squares, Error | 5.5 |
The test statistic for testing the hypothesis that mean of three treatments is equal is given by,
Using the calculated values we have, F= 51.21, the tabulated F at 0.05 level of significance is = 3.89.
As tabulated F < F, we reject the null hypothesis.
Also the p value for above statistic is <0.05, which also suggests no strong evidence against the null hypothesis. Thus we can conclude that Mean of three treatments are different.
b) When there are equal no. of observations in each treatment Fisher's LSD between any two treatment is defined as
, here d if d.f. of Error, a is level of significance and MSE is Sum of Square, Error.
Here d= 12 and a=0.05 then two tailed t (12,0.05) = 2.18
Using above data LSD = 11.201
Difference (mean) | Absolute Value | Conclusion |
(A-B) = -15 | 15 | As AbsVal > LSD, we conclude there is significant diff between treatment A & B |
(A-C) = -7 | 7 | As AbsVal < LSD, we conclude there is no significant diff between treatment A & B |
(B-C) = 8 | 8 | As AbsVal < LSD, we conclude there is no significant diff between treatment A & B |
c)