Adopting Online Learning in College Developmental Education Coursework: Impact on Course Persistence, Completion, and Subsequent Success

ABSTRACT

Online learning has been increasing rapidly at community colleges, especially in low-division high-demand coursework, such as developmental education. While existing studies have identified negative effects associated with online instruction in semester-long coursework in this particular setting, there is less evidence about how distance learning influences students’ completion of developmental education coursework and, more importantly, their subsequent academic outcomes. This paper examines the impact of fully online instruction, compared with traditional face-to-face instruction, on both concurrent developmental course outcomes and downstream outcomes. We use an administrative dataset from a state community college system that includes longitudinal student-unit record data from more than 40,000 students enrolled in developmental education courses between 2005 and 2012, and employ a two-way fixed effects model that controls for selection both at the course- and student-level. We find that taking one’s first developmental course through the online format reduces developmental course completion rate by 13 percentage points and subsequent enrollment in the gatekeeper course by 7 percentage points. Successful completion of developmental and subsequent gateway coursework represents critical milestones among community college students. This paper provides insight on how delivery format influences both the concurrent and downstream outcomes of developmental education students.

Acknowledgments

The authors thank the state for providing high-quality data and expert guidance on the state context. The research reported here was jointly supported by the Institute of Education Sciences, U.S. Department of Education, through Grant R305C110011 to Teachers College, Columbia University, and by the National Science Foundation, through Grant 1750386 to University of California, Irvine. The opinions expressed are those of the authors and do not represent views of the U.S. Department of Education or the National Science Foundation.

This study has been determined as “exempt” from IRB review, as it is secondary research of archived de-identified data, where the identity of human subjects cannot readily be ascertained.

Notes

¹ Although developmental coursework is provided in multiple fields of study, it is most prevalent in English and math; for the subject of English, it is a typical practice for colleges to assess and prepare students in reading and writing separately. The traditional developmental program typically consists of a sequence of multiple courses across several semesters, and students are assigned to different combinations of course requirements based on their initial assessment test scores in the corresponding subject area upon initial college enrollment (Boatman & Long, 2018; Chen, 2016; Grubb, 2013). For example, in a three-course math sequence, the lowest level of developmental math usually covers arithmetic, the middle level covers beginning algebra concepts, and the highest level prepares students for college algebra.

² A more detailed review of the existing literature on developmental education and online teaching and learning at community colleges can be found at https://drive.google.com/file/d/1Ly2IbyXPDZ3w58QWQ-HT6B0nFmnu6-8U/view

³ Authors’ calculation based on ASCCS administrative data for entering cohorts from 2005–06 to 2008–09.

⁴ A detailed explanation of how students are placed into different levels of developmental education coursework can be found at https://drive.google.com/file/d/1Ly2IbyXPDZ3w58QWQ-HT6B0nFmnu6-8U/view.

⁵ “Ever-online students” are defined as those who had ever taken an online course.

⁶ Studies that use data from other states also indicate that instructor characteristics account for only a negligible portion of the performance decrement associated with online learning (e.g., Hart et al., 2018).

⁷ For binary outcome measures, we used a linear probability model instead of nonlinear models (such as logit or probit models) due to the methodological challenges associated with the incidental parameters problem with a panel data structure. Specifically, unlike linear regressions that have the property of generating unbiased estimators, logit or probit models generate consistent estimators, meaning that as the ratio of the number of observations to number of parameters increases, the parameter estimates will converge onto their true values as standard errors become arbitrarily small. However, the problem with using non-linear models with panel data – especially when they have a short panel as in the current case – is that the number of parameters grows with the number of observations. That is, we would only have a fixed number of observations to estimate the intercept for each individual and that would not improve when we add additional individuals to the data. As a result, the parameter estimates cannot converge to their true value as the sample size increases, which makes the FE estimator to be inconsistent.

⁸ Studies that use more recent data from other states indicate that the gateway enrollment rates have been fairly comparable across states and over time. For example, based data from students who enrolled in developmental math courses in their first semester between 2008–2009 and 2013–2014 in the Kentucky Community and Technical College System (KCTCS), the author found that the average probability of gatekeeper enrollment in math is 38% (Kozakowski, 2019).

Additional information

Funding

This work was supported by the Institute of Education Sciences [R305C110011]; National Science Foundation [1750386].