98 Sample Size Calculations for Clustered and Longitudinal Outcomes
Suppose in a clinical trial the proportions of subjects who would potentially
follow the IM and MM patterns are w and (1 −w), respectively. Then a more
general sample size formula to accommodate a mixture of missing patterns
can be obtained by replacing δ
j
and δ
jj
0
in (4.9) with δ
(MIX)
j
= wδ
(IM)
j
+
(1 − w)δ
(MM )
j
and δ
(MIX)
jj
0
= wδ
(IM)
jj
0
+ (1 − w)δ
(MM )
jj
0
. Here the superscript
(MIX)
indicates that the marginal and joint probabilities are calculated under
the mixture of missing patterns. It is easy to show that with all the other
designing parameters being the same, we have n
(IM)
≤ n
(MIX)
≤ n
(MM )
.
There are many other methods to account for missing data. Dawson [21]
presented a sample size ...