勝算、勝算比、相對風險、對數勝算之截取冪級數估計量

Analysis of binary data in a longitudinal study has been an important statistical issue. Thus odds, odds ratio, relative risk, and log odds have a significant effect upon the binary variable. According to Lehmann in 1983, the unbiased estimator of the inverse of proportion is nonexistent but we can find a estimative value approximated to unbiased, and therefore use Conventional Point Estimators ( CE ) and Truncated Power Series Estimators ( TPSE ) to estimate for odds, odds ratio, relative risk, and log odds. Proceed to the next step, we use statistical software-R to simulate them, then we can obtain the data of Bias and MSE to compare by CE and TPSE, and then we can judge pros and cons of the two estimative methods by MSE. Finally, we can realize the conclusion that the TPSE for odds, odds ratio, relative risk, and log odds are the best efficient estimators.