ABSTRACT

Throughout industry measuring, inspection and testing is being done every minute by people and by automatic devices. The results are in the form of numbers, e.g., measurements or a count of the number on nonconforming pieces in a sample. These are numbers or data. They are obtained in order to take action on a manufacturing process for improving and rectifying it or for a decision on incoming products. An ever present character of such data or results is that they vary: from time to time, piece to piece, sample to sample. And this is true even if the production process is held as constant as is humanly possible. The same can be said for administrative or service functions where the product is a service rather than a manufactured unit. The service delivered, e.g., maintenance, consultation, testing or project management, can be highly variable. It also can be evaluated against some defined criteria, standard or with respect to the delivery of similar or equivalent services. Wherever we have variation, we have a statistical problem, whether we know it or not. Thus in such problems, we ought to be using the very methods designed for analysis of data, i.e., statistical methods. They can be used to

minimize the chances of wrong decisions. Fortunately, there are many statistical techniques which can be readily learned by anyone with even quite limited background. They will greatly aid in process improvement and control and in decision making on lots of products or in the delivery of services.