Data Collectionmedium
0:00.0

In a study using a ratio estimator Y^R=yˉxˉX\hat{Y}_R = \frac{\bar{y}}{\bar{x}} X, the variance is approximated as V(Y^R)N21fn(Sy2+R2Sx22RSyx)V(\hat{Y}_R) \approx N^2 \frac{1-f}{n} (S_y^2 + R^2 S_x^2 - 2R S_{yx}). What is the condition under which the ratio estimator is more efficient than the SRS mean estimator Y^=Nyˉ\hat{Y} = N\bar{y}?