fa] ar 1 uth)s 4 ne | >| wo 3 a a] ia ot le At) g » a +l W The variance and confidence {interval are n /yY,? n f=1 Vy i= v(R,) = {% (=) - n n v,\7/ ¥ x}/cmn ox? i=l int w and R, + tW(R,), respectively, where te is defined above. i=l — Ratio Estimation When Var(¥) is Independent of X. FIGURE 4. Ratio Estimation When Var(Y) is Proportional to X . |. n » xX FICURE 3. os sim n En A third situation (Figure 5} is when the standard deviation of Y is proportional to X. The optimal estimator of the slope of the regression line (average ratio} is then i Its variance is estimated by n /¥, 2 V(R,) = {> (z i=l i fay 2 ae x } Joon), i=1 "i and its confidence interval by X 606 607