In: Statistics and Probability
Twenty-four subjects were used in a clinical trial in which three factors were applied in a completely randomized 23 arrangement with three replicates for each treatment. The three factors were A, B, and C. The response variable was measured at a subsequent stage and the data were:
B1 | B2 | |||
C1 | C2 | C1 | C2 | |
A1 | 29 33 30 | 40 40 44 | 63 59 56 | 60 68 58 |
A2 | 36 39 37 | 48 52 52 | 60 65 59 | 73 61 68 |
a) write the linear model for the experiment, explain terms, and computer the analysis of variance for the data
b) Prepare a table of cell and marginal means with their respective standard errors.
c) Test the null hypothesis for all main and interaction effects.
d) Comment in detail on the difference between treatment means.
a)
b)
sample size | marginal means | varince | standard errors | ||
A | A1 | 12 | 48.33 | 191.52 | 3.99 |
A2 | 12 | 55.00 | 168.91 | 3.75 | |
B | B1 | 12 | 40.00 | 60.36 | 2.24 |
B2 | 12 | 63.33 | 27.33 | 1.51 | |
C | C1 | 12 | 48.00 | 230.91 | 4.39 |
C2 | 12 | 55.33 | 124.42 | 3.22 |
c)
Tests of Between-Subjects Effects | |||||
Dependent Variable:Y | |||||
Source | Type III Sum of Squares | df | Mean Square | F | Sig. |
Corrected Model | 4002.667a | 7 | 571.810 | 40.010 | .000 |
Intercept | 64066.667 | 1 | 64066.667 | 4482.799 | .000 |
A | 266.667 | 1 | 266.667 | 18.659 | .001 |
B | 3266.667 | 1 | 3266.667 | 228.571 | .000 |
C | 322.667 | 1 | 322.667 | 22.577 | .000 |
A * B | 10.667 | 1 | 10.667 | .746 | .400 |
A * C | 2.667 | 1 | 2.667 | .187 | .672 |
B * C | 130.667 | 1 | 130.667 | 9.143 | .008 |
A * B * C | 2.667 | 1 | 2.667 | .187 | .672 |
Error | 228.667 | 16 | 14.292 | ||
Total | 68298.000 | 24 | |||
Corrected Total | 4231.333 | 23 |
Comment: The p-value of the interaction A, B, and C is 0.672. Hence, this interaction does not a significant effect on the model. Among the interaction of the two treatments, the interaction of the treatments B and C is 0.008. Hence, the interaction of this has a significant effect on the model. We know that if the higher order interaction of the treatments is significant than the effect of the lower order or main effect do not the meaning. Hence, the effect of the main effect in this model does not the meaning.
d)
Here, the interaction has a significant effect. Hence, the difference between treatment means do not have meaningful.