“Big Data” in Educational Administration: An Application for Predicting School Dropout Risk

Abstract

Purpose: In an era of unprecedented student measurement and emphasis on data-driven educational decision making, the full potential for using data to target resources to students has yet to be realized. This study explores the utility of machine-learning techniques with large-scale administrative data to identify student dropout risk. Research Methods: Using longitudinal student records data from the North Carolina Department of Public Instruction, this article assesses modern prediction techniques, with a focus on tree-based classification methods and support vector machines. These methods incorporate 74 predictors measures from Grades 3 through 8, including academic achievement, behavioral indicators, and socioeconomic and demographic characteristics. Findings: Two of the assessed classification algorithms predict high school graduation and dropping out correctly for more than 90% of an out-of-sample student cohort. Findings reveal a shift toward lower dropout incidence in regions hit hardest by the economic recession of 2008, especially for male students. Implications for Research and Practice: Machine-learning procedures, as demonstrated in this study, offer promise for allowing administrators to reliably identify students at risk of dropping out of school so as to provide targeted, intensive programs at the lowest possible cost.

Predicting Students' Attitudes Toward Collaboration: Evidence From Structural Equation Model Trees and Forests

“…We believe that administrators and researchers can benefit from data mining techniques to glean formerly “invisible” information about students and to greatly augment existing understandings of educational phenomena. What's more, its interpretable tree structure rules are useful for setting up early warning systems, especially in the fields of “high-risk” topics like school dropout (Sorensen, 2019 ) and self-injury (Ammerman et al, 2020 ).…”

“…In this study, independent variables were intentionally classified into two primary domains: demographic variables and variables of other attitudes and activities. This was based on the following considerations: (1) demographic characteristics of students are easily available for educators, which means such variables are the most fundamental to some extent; (2) other attitudes and activities of students may reflect their deeper traits and may therefore be more predictive, but they are not convenient to obtain; and (3) mixing the two types of characteristics may cause demographic characteristics to appear less salient (Sorensen, 2019 ). Therefore, we analyzed the two types of datasets separately: (1) dataset with demographic variables and (2) dataset with demographic variables as well as variables of other attitudes and activities (i.e., the whole dataset).…”

“…The development of information technology provides a convenient way for educators to collect data on students' development, while the data mining techniques improve the use of big data in education (Sin and Muthu, 2015 ). For example, Sorensen ( 2019 ) used machine learning techniques incorporating 74 predictors about academic achievement, behavioral indicators, and socioeconomic and demographic characteristics to identify student dropout risk. In the context of DDDM, the current study used data mining techniques to analyze a large scale dataset about various aspects of students, aiming to identify which factors are most important for students' attitudes toward collaboration and reveal their interactions.…”

“…In the context of DDDM, the current study used data mining techniques to analyze a large scale dataset about various aspects of students, aiming to identify which factors are most important for students' attitudes toward collaboration and reveal their interactions. For educators, data mining from a large number of potential predictors may provide a complementary way, which is beneficial for educators to focus on key factors and make effective interventions at a lower cost (Sorensen, 2019 ).…”

Dashboard stories: How narratives told by predictive analytics reconfigure roles, risk and sociality in education

“…Qazdar et al, 2019) or (3) academic, social and demographic data, i.e., including gendered/racialised identity markers, age and socioeconomic indicators like free school meals (e.g. Sorensen, 2019).…”