Abstract: In this paper, we develop a duality of sample size
calculation given the odds ratio and the risk difference. Given the odds ratio,
the sample size attains minimum if and only if the risks of two groups are
symmetric aboutwhich is equivalent to
that the risk difference attains maximum. On the other hand, given risk
difference, the sample size attains maximum if and only if the risks of two
groups are symmetric about which is equivalent to that the odds
ratio attains minimum. We also obtain rules of thumb to calculate sample only
given the odds ratio or risk difference for a relative range of risks. The
method of this paper can be widely in epidemiology and health research.
Keywords and phrases: risk ratio, risk difference, epidemiology study, categorical data analysis.