While there has been a great deal of research into how to increase the number of responses to a mail survey, very little has been done on the question of how many surveys should be mailed out.; The first part of this work deals with how to model the survey response pattern. This is then used as a foundation for a two stage optimization problem to determine the optimal number of questionnaires.; Past experience and tuition are included by means of a prior distribution. A confidence constraint is imposed whereby there is a prespecified probability of meeting some desired level of returns by a specified deadline. The resulting problem can be solved using numerical integration and search routines. Simulations are performed to compare this procedure with the previous work of Parasuraman (1982).; The last part of this work introduces a new regression model, the Odds Ratio Model, for problems with a binary response variable. This provides a conceptually attractive alternative to the logistic regression model. |