home - login - register

PERC 2009 Abstract Detail Page

Previous Page  |  New Search  |  Browse All

Abstract Title: Time-Series Analysis: Detecting & Measuring Structural Changes in Knowledge
Abstract: Time-series designs are an alternative to pretest-posttest methods that are able to identify and measure the impact of multiple educational interventions.  In this study, we use an instrument employing standard multiple-choice conceptual questions (e.g., from CSEM) to collect data from students at regular intervals.  The questions are modified by asking students to distribute 100 'confidence points' among the options in order to indicate which options they think are more likely to be correct.  Tracking the class-averaged confidence ratings for each option produces a set of time-series.  Intervention ARIMA (autoregressive integrated moving average) analysis can then be used to test for, and measure, structural changes in each series.  In particular, it is possible to discern which interventions (lectures, labs, etc.) were able to produce significant and sustained changes in class performance.  Cluster analysis can also identify groups of students whose ratings evolve in similar ways.  Methods and preliminary findings are presented.
Abstract Type: Contributed Poster
Contributed Poster: Download the Contributed Poster

Author/Organizer Information

Primary Contact: Aaron R. Warren
Purdue Unversity North Central
1401 S. US-421
Westville, IN 46391
Phone: 219-785-5659
Fax: 219-785-5507