Interactive applets on the Web for methods and statistics

Abstract

We present two web applications designed to optimally fulfill the criteria of parsimony, usability, and availability (platform independence) in the teaching of important concepts in research methods and statistics. Effects of confounded variables may bias results in within-subjects designs, and can be avoided or exaggerated by counterbalancing of order. The first web application seeks to illustrate underlying concepts and relationships between true effects, confound effects, and choice of counterbalancing schemes by letting the user continuously manipulate the relevant variables. In the second web application regression is conceptualized interactively by dragging the regression line to minimize continuously updated squares together representing the sum of squared errors. The sites can be viewed at http://www.bolderstats.com/within/index.html and http://www.bolderstats.com/regfit/index.html