The purpose of this article was (1) to review the two
major approaches to sample size estimation for paired binary
data and describe how they differ; and 2) to compare
the sample size estimates based on a McNemar test vs.
GEE on the same numerical examples. In this article,we suggest that in the special case of a 2 2 table, the
sample size estimation method need not align with the
planned analysis. An asymptotic unconditional McNemar
test can be used to estimate sample size irrespective of
the planned analysis.