Using Accessible Middle School Data to Predict High School Success

Renee HoardFollow

Date

3-2016

Department

School of Education

Degree

Doctor of Education (EdD)

Chair

Scott Watson

Keywords

dropout, graduation, high school, multiple step-wise regression

Disciplines

Curriculum and Instruction | Curriculum and Social Inquiry | Education | Educational Assessment, Evaluation, and Research | Educational Methods | Other Education

Abstract

New federal adequate yearly progress accountability laws require that schools successfully graduate students within 4 years. The purpose of this quantitative, multiple step- wise regression analysis study is to identify the most parsimonious set of predictor values to predict successful graduation from high school within four years, using data that is readily available on existing school data systems. This study has identified ten research-based predictor variables (independent) and one criterion variable (dependent). The dependent variable is defined as successful graduation from high school with 28 or more credits within four years. The independent variables include number of days absent, grade point average, number of failed courses in middle school, number of "did not meet" CRCT tests in middle school, gender, ethnicity, socioeconomic status, number of times suspended, number of math classes failed, and age. These were selected as the initial variables for consideration in constructing this multiple regression model because these data are archival and easily obtained from most current school databases. The multiple step-wise regression analysis ranked each predictor variable by contribution, allowing researchers to observe which variables most significantly impact success. Analysis found that grade point average accounts for approximately 40% of the variance in whether or not a student successfully graduates from high school with 28 credits within the standard four years. Further, GPA combined with "did not meet" on middle school criterion referenced competency tests added an additional 5% change to the model, accounting for 45% of the variance.

This document is currently not available here.

Share

COinS