Report

R. Webster West Integrated Analytics LLC Texas A&M University Applets abound Applets are small Web-based interactive programs written typically in Java or Flash. To steal a phrase from Apple, “Pick a statistical topic and there is an applet for that.” There is a great deal of variation in terms of quality and in terms of how applets are used within courses, but a fair question to ask is: “What does the average student really get out of using applets?” A confidence interval applet http://www.stat.tamu.edu/~west/ph/meanci.html Questions about this applet Do students really understand how to manipulate the underlying data structure using the applet’s widgets? How much instructor guidance is required? Can students use the applet on their own? Should the applet presentation be followed with an exercise? Can students comprehend the large number of educational objectives targeted by the applet? Other approaches Would a programmatic simulation written in a language like R be a better way of teaching these concepts? Probably not. The typical student has no ability to digest computer code, and they certainly can not write their own code. A few simple additions to the StatCrunch package allows for a middle ground approach between the mindless clicking and complex programming alternatives. The StatCrunch approach Almost all StatCrunch procedures offer a Group By option which makes it simple to compute/compare results for different groups. Two newer ingredients: The ability to split and stack simulated data. The ability to save results from numerous procedures to the data table so that further computations are possible. Add these items together and you have a powerful recipe for doing pedagogical simulations in StatCrunch. Today’s StatCrunch demo A confidence interval simulation in StatCrunch A short glimpse at future capabilities Other possible StatCrunch simulations: Sampling distributions and the CLT Hypothesis testing Sampling distributions of regression estimates Resampling