Repeated simulated sampling in excel as a tool for teaching statistics

With the growth in the availability of inexpensive computing the emphasis in teaching introductory statistics must shift from the mechanics of performing statistical procedure n a calculator or by hand to the interpretation of results easily obtained by computer. Textbooks in statistics provide ample instruction on using technology to perform statistical analysis but provide little in the way of hands on activity or simulation to teach abstract concepts. Many sites on the World Wide Web have Java applets that allow users to simulate sampling but these are subject to change both in how they work and where they are located which makes planning instruction with these programs difficult. Several software programs have been specially developed to allow for simulation of sampling distribution but some colleges may be unwilling or unable to purchase or support such specialized packages. In spite of lacking the graphics of commercial packages and web-based simulations, simulation exercises in Excel have the advantage of utilizing a program that is available and supported on most campuses.<br>In this paper I describe two exercises which use repeated simulated sampling in Microsoft Excel to teach about sampling distributions, particularly the Central Limit Theorem and the interpretation of confidence intervals. First, I will give a brief overview of the key concepts in the Central Limit Theorem and confidence interval estimation, describe the difficulties students have in learning these concepts and describe the potential of simulation to add clarity to these concepts. Next, I will describe how Excel is used to simulate sampling in my courses. Next I will describe the assignments I use in my classes to teach Central Limit Theorem and confidence interval estimation. Finally I will discuss possible exercises in repeated simulated sampling to teach other concepts in statistical inference as well as ways in which the effectiveness of these simulations in promotion g student learning can be formally evaluated.