Degree Level





Schools across the United States and throughout the world administer tests to students to evaluate their academic performance. In many instances, however, especially in classrooms with higher populations of racial/ethnic minorities and low SES students, there are often missing scores, attributed to higher rates of absenteeism among these demographics (Callahan, 2019; Friedman-Krauss & Raver, 2015; Evans, 2004). When evaluating within subjects, longitudinal data, many will utilize a pre, post significance test such as a paired samples T test or a Repeated Measures ANOVA, however, due to the assumptions of these tests, missing data has posed a problem and requires data manipulation tactics that may distort data representation. Unfortunately, data misrepresentation may disproportionately affect students of the described demographic. Prior studies have explored the possibility of utilizing independent-samples T tests and One Way ANOVAs as a method of significance testing and have found that in analyzing data containing missingness, these tests yield less biased results with higher amounts of statistical power. The current paper continues on this path, exploring the extent to which One Way ANOVAs exhibit results with higher statistical power as it relates to mean difference values in skewed data containing missingness, as compared to Repeated Measures ANOVAs. Although only simulated data was used, it was found that One Way ANOVAs outperformed Repeated Measures ANOVAs and as evidenced by results with lower rates of type I and type II error.

Publication Date

Spring 3-25-2022

Document Type