ST4241: Design and Analysis of Clinical Trials - The Department of ...
ST4241: Design and Analysis of Clinical Trials - The Department of ...
ST4241: Design and Analysis of Clinical Trials - The Department of ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
a2 a3 a4 a5 a6<br />
1.750000 1.083333 3.333333 3.416667 1.416667<br />
a2 a3 a4 a5 a6<br />
a2 4.640741 2.320370 2.320370 2.320370 2.320370<br />
a3 2.320370 4.640741 2.320370 2.320370 2.320370<br />
a4 2.320370 2.320370 4.640741 2.320370 2.320370<br />
a5 2.320370 2.320370 2.320370 4.640741 2.320370<br />
a6 2.320370 2.320370 2.320370 2.320370 4.640741<br />
By using the above information, the test statistics for the three contrasts are computed as<br />
L 1 = −0.6633, L 2 = −1.1010, L 3 = 0.0734.<br />
<strong>The</strong> Schiffe criterion should be used. <strong>The</strong> critical value is given by √ 5F 5,15,0.05 = 3.8087.<br />
None <strong>of</strong> the contrasts is significant.<br />
3. <strong>The</strong> data form the study comparing four formulations considered in the lecture notes is given<br />
at the end <strong>of</strong> this question. Using R, fit a linear model to the data. Using the information in<br />
the fitted object, do the following:<br />
(i) By computing the means for each formulation, it is obvious form that the estimated effect<br />
<strong>of</strong> Formulation 3 is different from all the others. Check that the value <strong>of</strong> the contrast<br />
C = a 3 − (a 1 + a 2 + a 4 )/3 is −0.6489 with an estimated st<strong>and</strong>ard error 0.0883. Test the<br />
significance <strong>of</strong> the contrast at level 0.05 using Scheffe’s criterion. (Why should Scheffe’s<br />
criterion be used?)<br />
<strong>The</strong> following code is used to extract the information from the fitted object lm.fit <strong>and</strong><br />
compute the required quantities:<br />
b=lm.fit$coef[14:16]<br />
v=vcov(lm.fit)[14:16,14:16]<br />
c1 = c(-1/3,1,-1/3)<br />
t(c1)%*%b<br />
sqrt( t(c1)%*%v%*%c1 )<br />
L = t(c1)%*%b/sqrt( t(c1)%*%v%*%c1 )<br />
<strong>The</strong> computation yields the given values for C <strong>and</strong> its st<strong>and</strong>ard error. <strong>The</strong> test statistic<br />
equals −7.35599. Its absolute value, compared with √ 3F 3,8,0.05 = 3.492641, is not<br />
significant.<br />
6