Relationship between test security policies and test score manipulations.

摘要

Published news reports, surveys, and previous research indicate that some educators have responded to the current high-stakes testing environment by manipulating test scores. This study first classifies the methods educators have used to manipulate test scores and then explores possible reasons why educators manipulate test scores. Next, possible methods to deter manipulations are explored, including the development and implementation of high quality test security policies. It is found that effective test security policies have four main components. These components are used to evaluate the test security policies currently implemented by all 50 states. These evaluations result in recommendations for improvements in test security policies.;After discussing plausible hypotheses for these score trend discrepancies, this dissertation examines the relationship between the quality of state test security policies and these scale-invariant score trend discrepancy effect sizes. No significant relationships were found for any of the time periods, subjects, or grade levels. Longitudinal analyses and group mean comparisons were also unable to find any significant relationships between test security policy quality and score trend discrepancies.;As a possible indicator of test score manipulations, comparisons are made between state test score trends and NAEP trends in reading and mathematics in grades 4 and 8. Due to technical limitations in comparing simple changes in proficiency rates, a scale-invariant framework based on P-P plots is used to estimate state-NAEP score trend discrepancy effect sizes. These effect sizes show that state trends were significantly larger than NAEP trends from 2003-05, 2005-07, and 2003-07.