nextupprevious
Next:About this document ... Up:Resampling Previous:Tests of Hypotheses

Estimation of Effect

Effect, $\Delta$, is estimated using three different statistics : mean, median, and Wilcoxon. Accordingly, we have three CC's which will perform each estimation. These CC's give the estimate, p-value, and 95% CI for the estimate. The CC's are As an example consider the above X-Y data set and corresponding output. We can see that $\hat{\Delta} = 4.0$ and the 95% CI for $\Delta$ is (0.0, 7.0).

When we have paired (dependent) samples the following CC gives results equivalent to the above.

As an example consider the following data. Assume the data is observed on twins who received two different treatments.
    Y : 9   12   10   15   12
    X : 7    5   14    6    8  
 Diff : 2    7   -4    9    4
Suppose we are testing H0 : X = Y versus HA : X < Y. The output given by the CC is
As we can see, $\hat{\Delta} = 4.0$ and the 95% CI for $\Delta$ is (-1.0, 8.0).


2000-08-21