12.01.2013 Views

ST 520 Statistical Principles of Clinical Trials - NCSU Statistics ...

ST 520 Statistical Principles of Clinical Trials - NCSU Statistics ...

ST 520 Statistical Principles of Clinical Trials - NCSU Statistics ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

CHAPTER 3 <strong>ST</strong> <strong>520</strong>, A. TSIATIS and D. Zhang<br />

� �<br />

πi(1 − πi)<br />

= E ,<br />

this means that we can obtain an unbiased estimator for E � �<br />

πi(1−πi)<br />

by using<br />

N −1<br />

N�<br />

i=1<br />

Summarizing these results, we have shown that<br />

• s 2 p =<br />

� N<br />

(pi−¯p) 2<br />

i=1<br />

N−1<br />

• We have also shown that N −1 � N i=1<br />

ni<br />

pi(1 − pi)<br />

ni − 1 .<br />

is an unbiased estimator for var(pi) which by (3.4) equals<br />

E<br />

� �<br />

πi(1 − πi)<br />

ni<br />

+ σ 2 π<br />

pi(1−pi)<br />

is an unbiased estimator for<br />

ni−1<br />

�<br />

E<br />

� πi(1 − πi)<br />

Consequently, by subtraction, we get that the estimator<br />

ˆσ 2 π =<br />

��Ni=1(pi − ¯p) 2<br />

� �<br />

− N<br />

N − 1<br />

−1<br />

N�<br />

i=1<br />

is an unbiased estimator for σ 2 π.<br />

ni<br />

ni<br />

�<br />

pi(1 − pi)<br />

ni − 1<br />

Going back to the example given in Table 3.1, we obtain the following:<br />

•<br />

•<br />

• Hence<br />

N −1<br />

� Ni=1(pi − ¯p) 2<br />

N − 1<br />

N�<br />

i=1<br />

pi(1 − pi)<br />

ni − 1<br />

= .0496<br />

= .0061<br />

ˆσ 2 π = .0496 − .0061 = .0435<br />

Thus the estimate for study to study standard deviation in the probability <strong>of</strong> response is given<br />

by<br />

ˆσπ = √ .0435 = .21.<br />

This is an enormous variation clearly indicating substantial study to study variation.<br />

PAGE 48

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!