Abstract
For a non-normal quality characteristic X, the sampling inspection plans are sometimes chosen from the plans based on the nomal distribution, or derived from approximating the X^^-'s distribution by the normal. In this paper, the effects of these normal approximations on characteristics of the plans for varizbles sampling inspection are evaluted by the aid of the Fisher and Comish's asymptotic expansion. The effects are measured by the differences of the producer's and the consumer's risks from the specified values. Then, it is shown that these risks greatly differ from the specified values when the X's distribution is regarded as the normal, and that on the other hand, the approximation is quite good when the plans are derived from the normal approximation to X^^-'s distribution.
