Examining Statistical Differences Between Constructed Response and Computer-Scorable Questions on Accelerated Math Placement Assessments

Date of Award

Fall 2023

Document Type

Dissertation

Degree Name

Doctor of Education in Organizational Leadership (EdD)

Committee Chair

Valerie Storey

Committee Member

Kevin Daberkow

Committee Member

Courtney Mckim

Abstract

The purpose of this study was to determine if there were significant statistical differences between scores on constructed response and computer-scorable questions on an accelerated middle school math placement test in a large urban school district in Ohio, and to ensure that all students have an opportunity to take the test. Five questions on a placement test given in the spring of 2018 were changed into questions that could be graded by computer for students taking the test in the spring of 2021. The data was examined for each test first using whole group data. Significant statistical differences were found in each of the five questions, with four questions having a higher mean as a constructed response item and one having a higher mean as a computer-scorable item. No significant differences were found when scores were analyzed by gender, and only one question had statistical difference when analyzed by ethnicity. Correlation was calculated between each student’s placement test score and their Math 8 state test score. Similar correlations were found between the two assessments during both testing years. To ensure that all students are provided an opportunity to succeed in an accelerated math pathway, all must need to be given access to helpful pre-testing materials and encouraged to take the placement test. Recommendations were made to continue using computer-scorable questions to provide an opportunity to identify as many qualified candidates for acceleration as possible.

Link to Full Text

Share

COinS