Using Bootstrapping to Measure the Effectiveness of Implementing an Immersion Method of Inquiry-Based Elementary Science Instruction: A Cluster-Randomized Trial

• Activity Number: 302
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Social Statistics Section
• Author(s): Mack Shelley*+ and Christopher Gonwa-Reeves and Marcia Laugerman and Luke Fostvedt and Brian Hand and William Therrien
• Companies: Iowa State University and Iowa State University and University of Iowa and Iowa State University and University of Iowa and University of Iowa
• Keywords: Bootstrapping ; mixed linear models ; Science Writing Heuristic ; Cornell Critical Thinking Test

Abstract

We examine the impact of implementation of the Science Writing Heuristic approach in elementary schools in Iowa as measured by Cornell Critical Thinking student test scores. The 48 schools were assigned to treatment or control using an R randomization algorithm based on percentage of students on free/reduced lunch, total enrollment, private or public status, and clustering within district. A linear mixed effects model was estimated for individual student (Level 1) and school/building (Level 2) using restricted maximum likelihood. Incorporating our mixed model with and without treatment and collapsing the residuals over the unit of randomization (school buildings) permits calculation of the true difference between treatment and control. This process was repeated 10,000 times, each time reassigning the treatment based on our randomization and calculating a new difference based on the new randomization. Building-level effect sizes demonstrate the considerably greater benefit received by students in the treatment group compared to the control group. Furthermore, these gains are substantial and sustained across total scores, scores on induction, and scores on deduction.