Comparison of Learning Outcomes from Traditional and Randomization-Based Inference Curricula in a Designed Experiment

• Activity Number: 651
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Education
• Author(s): Karsten Maurer*+ and Dennis Lock
• Companies: Iowa State University and Iowa State University
• Keywords: randomization based inference ; introductory statistics eduction ; curriculum study ; randomized assignment ; learning outcomes ; technology

Abstract

Within an introductory statistics education the topic of statistical inference is a crucial component of the course curriculum, normally spending the better part of a semester learning to draw conclusions about a population from data. These topics have traditionally been introduced by first teaching the normal distribution, followed by the concepts of how inference may be conducted and interpreted within this framework of normal theory. In recent years a randomization based approach to inference education has been advocated as an alternative. This method first utilizes computer simulation to drive a conceptual understanding of inference topics, followed by how inference can be conducted using normal approximation. A classroom study was conducted to compare the efficacy of these two curriculum options. The study employed a co-teaching structure and strategic room scheduling to allow for randomization of students to curriculum treatments. Learning outcomes on inference topics were then assessed using ARTIST scaled questions.