Sample Size Calculation Based on the Inference of Relative Risk and Odds Ratio

J Health Info Stat > Volume 36(1); 2011 > Article

J Health Info Stat 2011;36(1):109-122.

두 모집단 비율 비교에서 상대위험도와 승산비에 기초한 표본크기 추정

조성경 , 강위창 , 채성산

Sample Size Calculation Based on the Inference of Relative Risk and Odds Ratio

Seong Kyung Cho , Wee Chang Kang , Seong San Chae

ABSTRACT

Objectives:
We suggested a method of calculating sample size to compare the relative risk or the odds ratio, in which we used the asymptotic distribution of the logarithm of its estimator. The empirical power of the proposed sample size was investigated through simulation studies.
Methods:
The proposed formula of calculating sample size was derived based on the score test statistic induced by the asymptotic distribution of the logarithm of the estimated relative risk or the estimated odds ratio. Through some simulation studies the empirical power of the proposed sample size was compared with that of the sample size formulated from the asymptotic distribution of the difference between two estimated proportions.
Results:
When true relative risks or odds ratios were greater than 1, the power of the proposed sample size was more close to the intended power than that of the sample size obtained by the commonly used formula in the comparison of two proportions.
Conclusions:
When the comparison of two proportions should be carried out with the relative risk or odds ratio, the proposed formulas gave us more valid sample size than the formula derived by the test statistic of the difference between two proportions.

Key words: Sample size, Comparison of proportions, Risk ratio, Odds ratio