Activity sampling procedure updated so it does what is says on the tin

A few years ago I was grading my MBA exams. One of the questions was on Activity Sampling which I have taught based on the the Wald Interval, the standard Operations Management textbook method for identifying the confidence interval. In the middle of the pile of scripts I started to wonder why the Wald Interval, which is based on the normal distribution, would work on the binomial problem within the Activity Sampling procedure? After the long hours grading, I started to research this issue. The precise problem is called the Binomial Proportion Confidence Interval and there are many approximate solutions, one of them being the Wald Interval. This Wald interval is easy to present and easy to use, but the results it provides are wrong 80% of the time! There is an exact solution to the problem that, historically was rather hard to use, is now easily implemented in Excel or by online maths resources such as Wolfram’s Alpha via the Inverse Beta distribution. These musings led to a rather unusual research article (for me at least), Disney, (2016). I have also created an R-shiny web app which can determine the confidence interval and sample size requirements directly from your activity sampling results. Needless-to-say, I have also updated the way I teach Activity Sampling!