“VisualRegression – A Tool to Explain the Link Between Parameters and Estimates in OLS”

Co-presenters:

Reidar Hagtvedt

Terry College of Business

University of Georgia

            289C Brooks Hall

Athens, GA 30602-6253

            Phone:   (706) 542-3889

Fax:     (706) 542-9434

Gregory Todd Jones, Georgia State University

Kari Jones, University of Georgia

Abstract:

Students often have trouble understanding the link between parameters and estimates when they are faced with learning univariate regression for the first time. To help students understand that link, we created a visual tool using VBA in Excel that allows a student to choose the data-generating process and examine the results. The student chooses the intercept, slope, and x-range of the population regression line. Additionally, the student specifies the standard deviation of the noise term and the sample size from which the program generates a random sample. The population line and the sample are displayed in a graph.  Then the OLS line is fitted and displayed on the graph. The model parameters and the estimates are displayed so that students may directly compare the two.  Frequent replication while adjusting the input parameters will then help build students’ intuition regarding the role of: the sample size, the standard deviation of the error term, the slope of the population regression line, and the range of the x-values.  We include suggested choices for parameters in the directions. A live demo will be presented at the session.