BUS 215 Lecture Notes - Lecture 7: Binomial Distribution, Confidence Interval, American Statistical Association
4 Apr 2016
School
Department
Course
Professor
Document Summary
If you are summing probabilities over several x values, you might get lucky and have positive errors cancel negative ones (but don"t count on it). Accordingly, we chose to examine maximum absolute error in the entire array of x values. Instead of plotting the maximum error as a function of n and p (raff, p. 296) we focus on the rules of thumb. Table 1 shows the maximum absolute error for various values of p using the sample size recommended by rule 1 and rule 2. Binom. dist(x,n,p,0) norm. dist(x+. 5,mu,sigma,1) norm. dist(x 0. 5,mu,sigma,1) where mu = n*p and sigma = sqrt(n*p*(1 p)). Appendix - rule of 10 doane-seward applied statistics in business and economics 4e page 1. Recommended sample sizes and errors for two rules of thumb. 0. 01 for np 5 for np 10. It is easier to visualize this information if it is presented in a chart, as in figure 2.
