Logging In to Press On: An Examination of High School Dropout and Completion Among Students With Disabilities in Online Courses

Abstract

Students with disabilities (SWDs) continue to experience rates of high school dropout greater than students not receiving special education services. Furthermore, there is a persistent gap in the rates of high school completion among students with and without disabilities. While criticized for lowering standards and learning, online learning represents a plausible mechanism to both decrease dropout and increase high school completion among SWDs. Drawing on theoretical frameworks advanced by Dynarski et al. and Cavanaugh et al., the current study uses a nationally representative data panel to investigate the association between online coursetaking among SWDs and two dependent variables: high school dropout and completion. Results indicated that online coursetaking was associated with increased probabilities of high school completion among SWDs. Implications and policy recommendations are discussed.

Keywords

online learning dropout high school completion students with disabilities

According to the National Center for Education Statistics (NCES, 2017), roughly 8 of every 10 students completed high school on time and with a regular diploma in 2015–2016. This percentage represents an all-time high in the nation’s adjusted cohort graduation rate (ACGR; Kamenetz & Turner, 2016). In fact, the nation’s ACGR has climbed from 79% in 2011 to 84.1% in 2015–2016. Moreover, during this same period, the rate at which students dropped out of high school decreased from 7.4% in 2010 to 5.9% in 2015 (McFarland et al., 2017). This is welcome news; dropping out of high school is linked to a number of negative economic and social outcomes (Rumberger, 2011). Students who drop out of high school before graduation work and earn less (Rouse, 2007) and, as a consequence, contribute less to the national economy (Belfield & Levin, 2007). Students who drop out from high school also experience greater social disturbances including increased odds of incarceration (Sum, Khatiwada, & McLaughlin, 2009).

Of concern, however, is the fact that students with disabilities (SWDs) continue to experience reduced rates of high school completion and increased rates of high school dropout relative to students not receiving special education services (Stark & Noel, 2015). In the 2015–2016 academic year, while it was the case that 84.1% of the nation’s students earned a high school diploma, the rate of high school completion among SWDs was just 65.5%, a differential of 18.6 percentage points. In 2013–2014, SWDs graduated from high school at a rate of 19.2 percentage points below that of students without disabilities; in 2012–2013, the completion gap was 19.5 percentage points. A similar gap exists with regard to high school dropout. In 2013–2014, just 6.5% of students without disabilities dropped out of high school. In this same year, roughly 18% of students served under the Individuals with Disabilities Education Act dropped out of high school, a rate nearly 3 times that of students without disabilities (McFarland, Stark, & Cui, 2016). Although the proportions of SWDs graduating from and dropping out of high school have increased and decreased, respectively, the persistent gaps in educational outcomes among students with and without disabilities suggest the continued presence of educational inequity in the nation.

It is important to note several caveats with these statistics before moving forward. First, the increase in the high school graduation rate has not gone without criticism (e.g., Heckman & Lafontaine, 2010; Kamenetz & Turner, 2016). A number of policy experts contend that rather than being reflective of increased rates of successful learning, the rise in graduation rates is more a function of decreased state-level graduation requirements, increased offerings of alternative and less academically demanding diplomas and even misleading reporting (Carsen, 2016; Kamenetz & Turner, 2016). Second, measuring both high school graduation and high school dropout is a complex and difficult task (Heckman & Lafontaine, 2010; Rumberger, 2011). The ACGR, now the most widely used measure of high school completion, but certainly not the only measure, has been criticized for not factoring in delayed high school graduates and for unduly penalizing schools that serve high numbers of delayed graduates or returning school dropouts (Sublett & Rumberger, 2017). Regarding high school dropout, Rumberger (2011) argued that dropout is more a process than an event and accurately measuring the national dropout rate is, therefore, highly problematic. While many students permanently dropout from high school, many more students experience brief, though impactful dropout episodes or events which are significant enough to hinder their academic performance and/or delay graduation (Sparks, 2016). Regardless of these technical caveats, long-term trends across a variety of measures indicate persistent disparities in the rates of high school completion and dropout among students with and without disabilities (Stark & Noel, 2015), with SWDs far less likely to complete and far more likely to drop out from high school.

Reducing Dropout and Increasing Graduation

While a great number of interventions have sought to increase graduation and decrease dropout rates in the nation, just a few of these interventions have focused on SWDs. Wilkins and Huckabee (2014) identified over 500 academic journal articles reporting on dropout prevention strategies and found just 19 that either touched or focused exclusively on SWDs. Wilkins and Huckabee (2014) found that the most common interventions in this select body of literature involved increased mentoring, services, and support. These were targeted to specific disability areas and learning options, including class setting and exit options. Notably, Wilkins and Huckabee (2014) concluded from their assessment of the literature that SWDs benefit from increased flexibility in educational contexts. In particular, they noted that SWDs were more likely to graduate from high school in states that provided increased flexibility in terms of graduation requirements, exemption from exit exams, and the option of obtaining a diploma based on individualized education plan (IEP) completion. Wilkins and Huckabee (2014) suggested that policy makers consider providing increased flexibility for SWDs to reduce dropout and to increase the rate of obtaining a diploma.

Wilkins and Huckabee’s (2014) calls for increased flexibility compliment the conclusions of Dynarski et al. (2008), who—not specifically referring to SWDs—recommended, among other things, schools to “personalize the learning environment and instructional process” (p. 30) in order to stem school dropout. Dynarski et al. asserted that a more personalized learning environment promotes a sense of community and improves the learning climate by fostering more intimate and respectful relationships between teacher and students. Also, the authors noted that “a high degree of personalization allows schools to focus intensively on why students are having difficulty, and actively work to address sources of difficulty” (p. 31). Calls to combat school dropout and increase completion through more personalized and flexible learning options have been echoed by others as well (e.g., EDUCAUSE, International Association for K–12 Online Learning (iNACOL), National Dropout Prevention Center). And while there are a number of available tools to provide personalized and differentiated education to diverse students, it is clear from existing research that education technology and, in particular, online learning can provide students, including SWDs, with the flexible, personalized, and rigorous learning environments potentially conducive to stemming dropout and aiding completion (Gross, Tuchman, & Patrick, 2018; Patrick, Kennedy, & Powell, 2013).

What Is K–12 Online Learning?

J. F. Watson and Kalmon (2005) defined online learning as a subset of distance education “in which instruction and content are delivered primarily over the internet” (p. 127). This definition was later adopted by the iNACOL (2013) in their Online Learning Definitions Project, which sought to provide researchers and policy makers with a comprehensive “set of definitions related to online and blended learning in order to develop policy, practice, and an understanding of and within the field” (p. 2). Building off this work, as well as the work of the National Association for Charter School Authorizers, J. Watson, Murin, Vashaw, Gemin, and Rapp (2011) later distinguished online courses and programs according to their comprehensiveness, reach, type, location, delivery, operational control, type of instruction, grade level, teacher–student interaction, and student–student interaction. Most parsimoniously, J. Watson et al. (2011) defined online learning as
instruction via a web-based educational delivery system that includes software to provide a structured learning environment. [Online learning] enhances and expands educational opportunities and may be synchronous (communication in which participants interact in real time, such as online video) or asynchronous (communication that is separated by time, such as email or online discussion forums). [Online courses and coursework] may be accessed from multiple settings (in school and/or out of school buildings). Blended learning combines online learning with other modes of instructional delivery. (p. 8)
What is clear from J. Watson et al. (2011) and iNACOL is that “online learning” does not refer to one specific method of course delivery or instruction, but several. Individual teachers often have the leeway to operationalize their online courses, such that two same-subject teachers within the same school could have different online versions of the same class. Online courses can be fully online or blended with face-to-face (FtF) elements; instruction in online courses can be delivered in real time or delayed; students can supplement their high school curriculum with online courses or enroll full-time in state or district virtual schools. According to the Center on Online Learning and Students with Disabilities (COLSD, 2016), students may complete courses online to supplement their coursework either for credit recovery or to access courses not offered through their schools (such as Advanced Placement [AP] courses). According to Tindle, East, and Mellard (2017), the vast majority of high school students take just one or two supplemental courses online while attending a public school full-time. At least one commonality across all online/blended courses, however, is that they incorporate and/or rely on some degree of technological mediation beyond that which is traditionally used in typical classroom-based learning environments.

Online learning in K–12 learning contexts has grown tremendously in recent years (Barbour & Reeves, 2009; De la Varre, Irvin, Jordan, Hannum, & Farmer, 2014; Gemin, Pape, Varshaw, & Watson, 2015). According to R. Smith, Clark, and Blomeyer (2005), educational inequities in access to education have driven the growth of online learning in K–12 schools. Rapid advances in educational technologies and the relative affordability of these technologies have also inspired growth in online learning. Changes to state, district, and local educational policies have also undoubtedly contributed to the recent growth in online learning; in fact, several states including Michigan and Florida now require students to pass a set number of online courses for graduation.

Less clear than the steady growth of online learning in K–12 education is the empirical research into the educational outcomes associated with learning online. On the one hand, critics of online learning have argued that the quality of instruction is much lower in K–12 online courses (Gabriel, 2011). Researchers have also found decreased engagement (Gill et al., 2015) as well as performance among students in online courses (Ahn & McEachin, 2017; Heissel, 2016). There is also a concern that schools and districts may “game the system” by using online credit recovery to push struggling or delayed students through low-quality online coursework (Picciano, Seaman, & Day, 2011). Yet, on the other hand, researchers have found that students in supplemental online courses perform as well or better as their FtF counterparts (Johnston & Barbour, 2013). Indeed, a number of comprehensive meta-analyses have concluded that students in online and FtF courses experience similar academic outcomes (Cavanaugh, Gillan, Kromrey, Hess, & Blomeyer, 2004; R. Smith, Clark, & Blomeyer, 2005; Tallent-Runnels et al., 2006). Dropout rates in online courses are higher than those in FtF courses (Carr, 2000; Dupin-Bryant, 2004; Wojciechowski & Palmer, 2005). However, despite the lower aggregate rates of persistence, school principals remain largely supportive of online learning because online coursework can increase school completion through credit recovery and supplemental instruction, provide a bridge from school to career, and allow for differentiated or more personalized instruction (Picciano & Seaman, 2010).

How Might Online Learning Help SWDs?

It is important to note that the majority of the extant empirical investigations of online learning in high school have not focused on SWDs. This is a concern. For SWDs, online coursework can be an impediment to learning when, for example, content is not accessible (Hashey & Stahl, 2014; S. J. Smith & Basham, 2014), when their accommodations are not met (Roberts, Crittenden, & Crittenden, 2011) or if a student has “major” visual, hearing, motor, and cognitive impairments (Crow, 2008). However, there remain a number of plausible arguments as to why SWDs might benefit from learning at a distance. These arguments relate to the claims advanced by Wilkins and Huckabee (2014) and Dynarski et al. () as well as organizations like iNACOL and COLSD who have argued for increased flexibility and more personalized learning as a means of increasing graduation and decreasing dropout. Indeed, a review of the existing literature suggests that one of the chief benefits of online learning is increased flexibility and the vast opportunities for personalized instruction (Belz & Müller-Hartman, 2003; Christensen, Horn, & Johnson, 2008; The Foundation for Blended and Online Learning and Evergreen Education Group, 2007). According to the Center for Online Learning and Students with Disabilities (2016), learning online offers a compelling alternative for SWDs because online learning provides “flexible scheduling; individual mentoring; safe communities in which to learn; and varied methods of teaching, curriculum delivery, and assessment” (p. 85).

Theoretical Framework

In light of the existing research into online learning among SWDs as well as the work of Wilkins and Huckabee (2014); Dynarski et al. (2008); Repetto, Cavanaugh, Wayer, and Liu (2010); and Cavanaugh, Repetto, Wayer, and Spitler (2013), the current study posits the following theoretical suppositions. First, online learning can provide SWDs with flexibility and personalization by giving them control over when they learn. In contrast to FtF learning contexts where students must adhere to a strict, preset schedule, learning online—perhaps from the comfort of home—can give SWDs the ability to read, watch, or listen to educational material when they are physically, cognitively, or emotionally best prepared and most supported to do so. This would be particularly true in asynchronous learning environments where students do not necessarily have real-time, prescheduled interactivity. Yet it seems that SWDs could gain a material degree of flexibility even in hybrid learning environments that may have a certain number of preset FTF meetings but also allow SWDS to complete their work at a distance. The hybrid learning format offers both the benefit of flexibility provided by online learning to suit a variety of students needs that might otherwise impede them in a traditional FtF setting, as well as the immediate support of an FtF learning environment (Snart, 2017).

Second, SWDs might gain additional flexibility and personalization through online learning by determining how they prefer to learn. An example is that FtF courses primarily rely on lecture-based instruction, often focused on a central textbook. In these traditional learning environments, students often sit for prolonged periods of time, in rows and in classrooms that may be distracting for students with cognitive disabilities. For students with mobility disabilities, traditional classroom learning environments may pose geospatial limitations. Learning online, on the other hand, while not entirely free of other constraints or limitations, is not as inflexible. Students can choose to listen, watch, or read their learning material, in solitude or the presence of a friendly adult advocate. SWDs experience physical restrictions as barrier to learning in a variety of ways, depending on their unique circumstances. Some forms of physical restrictions include inaccessibility of campus facilities and environmental factors such as allergens, lighting, and acoustics (Skårbrevik, 2005).

Third, online learning environments can provide SWDs with flexibility and personalization in the way they communicate interpersonally with their peers and teachers. For many SWDs, interacting in the traditional classroom setting can be intimidating and consequently be an obstacle to academic performance (Shah, 2011). Judgment from peers may heighten communication anxiety and, subsequently, stifle learning. Online environments, on the other hand, can allow students to communicate with their teachers through preferred and more comfortable communication media such as the phone, synchronous or asynchronous video, and e-mail or discussion forums. This interpersonal flexibility gives students increased control over their learning and may have the added benefit of increasing rather than decreasing engagement and climate, just as Dynarski et al. (2008) argued.

While administrators and teachers must work diligently to ensure their online programs and courses are student-centered and built around solid instructional practices (Burdette, Greer, Woods, & Kari, 2013), once these programs are established, it stands to reason that robust online course offerings can help SWDs complete and persist in high school. Cavanaugh et al. (2013) provided an illustrative framework for how this might happen in practice and it is this framework, combined with the work of Wilkins and Huckabee (2014) and Dynarski et al. (2008), that informs the current study. Specifically, Cavanaugh et al. (2013) argued that online learning can provide SWDs with five critical factors for decreasing dropout: increased learner control, a flexible and rigorous curriculum, a safe climate, a caring community, and connections to students as individuals and their future goals. These “5 Cs” of Student Engagement place into the context of SWDs the assertions of Wilkins and Huckabee (2014) and Dynarski et al. (2008). Unfortunately, too few studies have investigated the association between online course participation in high school and graduation and dropout among SWDs in order to verify this framework. Consequently, the current study seeks to address this gap and to build upon the literature by using nationally representative data to answer the following key research questions:
Research Question 1: What is the association between online course participation in high school among SWDs and experiencing a high school dropout episode.

Research Question 2: What is the association between online course participation in high school among SWDs and high school graduation?

Data and Method

Data for the current study came from the High School Longitudinal Study of 2009 (HSLS:09), a nationally representative, multiyear data collection effort by the NCES. Data for HSLS were collected over several waves spanning from the fall of 2009, when participants were in the ninth grade, to 2013 when most participants had graduated high school (Ingles et al., 2015). HSLS:09 contains a rich panel of data that touches on many aspects of high school students’ academic, personal, familial, behavioral, and attitudinal characteristics. These data are complemented with information gathered from parent, teacher, and administrator questionnaires. In addition, complete high school transcripts for participants were made available in restricted versions of the data. These transcript data provide detailed coursetaking information for student participants that allowed the authors of the current study to know the types of courses students completed in high school. In addition, these transcript data revealed whether students completed a given course in an online setting. While the entire HSLS sample approximates 25,000 distinct student-level observations, a relatively small number of the sample members had complete information pertaining to disability status (n = 11,000). As a result, the complete HSLS sample was reduced to just these cases. There were also a number of missing observations on other key variables of interest, and to preserve statistical power and to avoid potential estimation bias resulting from listwise deletion, we elected to impute missing values for participants in our analytic sample using multiple imputation (McCleary, 2002; Royston, 2004). In more detail, five sets of plausible values were estimated and then imputed back to the sample in cases in which NCES-provided sample weights were set to nonzero. These weights were used for imputation and during all empirical analyses. Throughout the current study, sample sizes have been rounded to the nearest 10s digit per NCES guidelines. After multiple imputation, the final analytic sample was composed of approximately 7,250 students from just under 700 schools. Table 1 provides the descriptive statistics for the students within this analytic sample.

Table 1.
Descriptive Statistics.

All Students No IEP IEP

Variables M SD M SD M SD

Outcome variables

Ever dropout 0.10 0.29 0.08 0.27 0.17 0.37

High school completer 0.92 0.28 0.93 0.25 0.84 0.37

Student demographics

Female 0.49 0.50 0.52 0.50 0.36 0.48

White 0.57 0.49 0.57 0.49 0.56 0.50

Black 0.08 0.28 0.07 0.26 0.13 0.34

Asian 0.08 0.27 0.09 0.28 0.04 0.19

Hispanic 0.17 0.37 0.17 0.37 0.17 0.38

Other 0.10 0.30 0.10 0.30 0.10 0.30

Socioeconomic status 0.08 0.80 0.14 0.79 −0.20 0.78

Academic characteristics

School engagement −0.32 2.03 −0.21 1.91 −0.78 2.43

School belonging 0.02 1.09 0.07 1.06 −0.19 1.17

High school GPA 2.78 0.80 2.87 0.77 2.39 0.79

Ratio of online credits to online courses 0.02 0.11 0.02 0.11 0.02 0.11

Friends who have dropped out 0.29 0.63 0.27 0.60 0.41 0.71

School characteristics

Public school 0.81 0.39 0.78 0.41 0.93 0.25

Catholic school 0.11 0.31 0.13 0.33 0.04 0.20

Other private school 0.08 0.27 0.09 0.29 0.03 0.17

City 0.29 0.45 0.30 0.46 0.23 0.42

Suburb 0.33 0.47 0.33 0.47 0.33 0.47

Town 0.13 0.34 0.14 0.34 0.13 0.34

Rural 0.25 0.43 0.23 0.42 0.31 0.46

Not a problem 0.33 0.47 0.36 0.48 0.21 0.41

Minor problem 0.43 0.50 0.42 0.49 0.48 0.50

Moderate problem 0.17 0.38 0.16 0.37 0.21 0.41

Serious problem 0.06 0.24 0.06 0.23 0.09 0.29

Dropout prevention program 0.39 0.49 0.38 0.49 0.46 0.50

Percent recieving services 11.36 8.02 10.77 7.76 13.85 8.57

Regular noncharter 0.95 0.22 0.95 0.21 0.94 0.24

Charter school 0.03 0.16 0.02 0.16 0.03 0.17

Special program or magnet 0.02 0.14 0.02 0.14 0.02 0.14

Vocational or tech school 0.00 0.06 0.00 0.04 0.01 0.09

Alternative school 0.00 0.03 0.00 0.02 0.00 0.06

Observations 7,250 5,870 1,380

Note. IEP = individualized education plan; GPA = grade point average.

Measures

There were two outcome variables of interest in the current study. The first outcome was a binary indicator of whether or not a student in the HSLS sample had at least one “dropout episode” during high school. A dropout episode was defined by NCES as an unexcused stoppage of high school attendance for a period of 4 weeks or more, not including summer or other school breaks. As Table 1 indicates, roughly 10% of the students in the analytic sample had experienced at least one such episode or event in high school. As Table 1 also indicates, SWDs experienced dropout events more frequently than students without disabilities at a rate of 2 to 1. This is in line with national statistics (McFarland et al., 2016).

The second outcome variable used in the current study was a binary indicator of high school completion. High school completion was defined as earning a traditional high school diploma, a GED, or similar equivalency by the spring of 2013, 4 years after the start of HSLS. Table 1 illustrates that of the students in the analytic sample, roughly 92% had successfully completed high school by the spring of 2013. Of the students in the sample with a formal IEP, the high school completion rate was approximately 84%. This is in contrast to the students in the sample without a known IEP, who as a group had a completion rate of approximately 93%, a 9-percentage point differential.

The main predictor variable for the current study was the interaction between the number of online courses a student took throughout high school and a binary indicator of whether that student had a formal IEP with his or her school by the start of the ninth grade. On one side of this, multiplicative term was the number of online courses students took in high school. Approximately 880 distinct students in the analytic sample chose to take at least one online course during high school. This represented approximately 8% of the sample. Table 2 lists the descriptive characteristics of these students. Figure 1 illustrates that the majority of online coursetakers took just a few courses online. More specifically, of the students who took high school courses online, 610 (or 70%) took just one or two online courses. Roughly 89% of online coursetakers took between 1 and 5 courses and 96% took between 1 and 10 courses. The range of online course enrollment among SWDs ran from 0 to 5 courses; for students not receiving special education services, the range ran from 0 to 35. Concerned that the variance in the online coursetaking predictor variable would be artificially inflated by a number of extreme cases, we first carried out our empirical analyses with all observations and, subsequently, excluding cases who exhibited high leverage statistics and Cook’s distances (seven cases) to compare the findings. Results were not sensitive to outliers.

Table 2.
Descriptive Statistics by Number of Online Courses.

Variables 1–2 Courses 3–5 Courses 6–10 Courses >10 Courses

M SD M SD M SD M SD

Outcome variables

Ever dropout 0.08 0.28 0.17 0.38 0.16 0.37 0.23 0.43

High school completer 0.95 0.23 0.88 0.33 0.93 0.26 0.81 0.40

Student demographics

Female 0.51 0.50 0.51 0.50 0.54 0.50 0.71 0.46

White 0.52 0.50 0.50 0.50 0.54 0.50 0.60 0.50

Black 0.11 0.31 0.09 0.29 0.18 0.39 0.09 0.28

Asian 0.09 0.29 0.08 0.27 0.08 0.28 0.00 0.00

Hispanic 0.16 0.37 0.18 0.38 0.13 0.34 0.20 0.41

Other 0.12 0.32 0.15 0.36 0.07 0.25 0.11 0.32

Socioeconomic status 0.11 0.83 0.04 0.74 0.10 0.78 −0.04 0.66

Academic characteristics

School engagement −0.78 2.61 −0.82 2.43 −0.75 2.45 −0.09 1.64

School belonging 0.10 0.95 −0.11 1.06 0.07 1.17 −0.03 1.20

High school GPA 2.72 0.84 2.48 0.89 2.65 0.94 2.47 0.76

Ratio of online credits to online courses 0.53 0.26 0.50 0.24 0.50 0.19 0.41 0.13

Friends who have dropped out 0.26 0.58 0.42 0.68 0.57 0.79 0.63 0.93

School characteristics

Public school 0.85 0.36 0.88 0.33 0.83 0.38 0.74 0.44

Catholic school 0.05 0.22 0.02 0.13 0.05 0.21 0.06 0.24

Other private school 0.10 0.30 0.11 0.31 0.13 0.34 0.20 0.41

City 0.30 0.46 0.28 0.45 0.30 0.46 0.54 0.51

Suburb 0.39 0.49 0.44 0.50 0.35 0.48 0.09 0.28

Town 0.07 0.25 0.06 0.24 0.06 0.25 0.09 0.28

Rural 0.24 0.43 0.21 0.41 0.29 0.46 0.29 0.46

Not a problem 0.30 0.46 0.29 0.45 0.32 0.47 0.36 0.49

Minor problem 0.46 0.50 0.46 0.50 0.53 0.50 0.42 0.50

Moderate problem 0.21 0.41 0.18 0.39 0.11 0.32 0.12 0.33

Serious problem 0.03 0.17 0.07 0.25 0.04 0.19 0.09 0.29

Dropout prevention program 0.44 0.50 0.36 0.48 0.39 0.49 0.36 0.49

Percent receiving services 11.75 7.66 12.41 7.18 11.77 9.16 11.50 8.56

Regular noncharter 0.95 0.22 0.91 0.29 0.91 0.29 0.94 0.24

Charter school 0.02 0.14 0.02 0.16 0.05 0.23 0.03 0.17

Special program or magnet 0.02 0.15 0.02 0.16 0.02 0.13 0.00 0.00

Vocational or tech school 0.00 0.06 0.00 0.00 0.00 0.00 0.00 0.00

Alternative school 0.01 0.08 0.04 0.20 0.02 0.13 0.03 0.17

Observations 607 176 63 35

Note. GPA = grade point average.

Figure 1.
HSLS Online Coursetaking.

It is imperative to note the inherent ambiguity regarding the operational definition of “online course” provided by HSLS. The HSLS transcript file was necessary to determine the number of online courses students completed throughout high school. Within this transcript file was a variable that specifies the location of every course a student attempted. Four locations were provided by NCES: high school, career/vocational center, college/university, and online. While we know how many courses students attempted and completed at an “online” location, what we cannot know from the available data was how and to what degree a given “online” course was technologically mediated. For example, the researchers were unable to know whether an “online” course was fully online or hybrid. And if a course was hybrid, the researchers were unable to confirm how many FtF meetings were required. Another aspect of the online courses students completed that the researchers did not know related the delivery format. Specifically, the researchers did not know whether an online course was delivered synchronously, asynchronously, or both. Thus, while the researchers were confident that courses completed at an “online” location were distinct and mutually exclusive from those completed in “high school,” the researchers were less confident about the exact nature of these courses. This represents a limitation of the current study.

On the other side of this multiplicative term was a binary indicator of whether a student in the HSLS sample began the ninth grade with a formal IEP with his or her school. There are legitimate criticisms with relying on a student’s IEP status a proxy for disability. First, not every student with a disability has an IEP. Some of these students may develop an IEP with their institutions in subsequent high school years. On the other hand, some of these students may never receive formal assistance and accommodations for the disability(ies) they have during high school. Second, one does not know which specific disability(ies) a student has. Indeed, the specific disabilities of the students with formal IEPs were not detailed in HSLS. HSLS researchers did ask parents in the base year of data collection if a doctor or school had told them their child had a special need or disability. Of the parents who responded to these questions, roughly 9% of parents said they were told by a doctor or school their child had a specific learning disability. Nearly 5% of parents were told their child had a developmental delay, just over 1% were told their child had some form of autism, nearly 3% had been told their child had a hearing or vision problem, nearly 3% had been told their child had a bone, muscle, or joint problem, less than 1% had been told their child had an intellectual disability, and just over 11% had been told their child had ADD or ADHD. These data offer some superficial insight into the range of potential services students in the sample received throughout HSLS. However, an inspection of the data revealed that not all students whose parents reported being told by a doctor or school that their child had one of the disabilities listed above actually had a formal IEP with their school. Consequently, these specific special needs/disability identifiers were not used in the analyses. This is another limitation of the current study.

To improve model estimation and to condition the relationship between online coursetaking and the outcome variables on a set of rich and theoretically appropriate student- and school-level factors, the researchers included a number of statistical controls that could be placed into the following three categories: “student demographics,” “academic characteristics,” and “school factors.” The first category, student demographics, included indicators for gender and race/ethnicity. The researchers were also sure to include a standardized measure of students’ socioeconomic status (SES). This particular measure was a composite variable constructed by NCES and derived using parent/guardian education, occupation, and family income.

The second set of control variables, academic characteristics, pertained to students’ academic achievement and involvement in school. The first variable in this set was an NCES-derived measure of school engagement. This measure was composed of the following individual Likert-scaled items: How often do you go to class without your homework complete, how often do you go to school without a pencil or paper, how often do you go to class without books, and how often do you go to class late? The school engagement measure was created using principal components analysis and subsequently standardized. NCES reported a reliability coefficient of α = .65 for this measure. The researchers also included in the set of academic characteristics a construct of school belonging that was based on the following items: How safe do you feel at this school, how proud are you of this school, whether adults are available to you to discuss problems with, the degree to which you feel school is a waste of time, and the degree to which you feel getting good grades is important. The NCES reported a reliability coefficient of α = .65 for this measure. To measure academic achievement, the researchers included each student’s cumulative high school grade point average. The researchers also included in this category of measures a count of the number of friends students reported having who had dropped out of high school. Last, in order to account for online course performance among students in the analytic sample, the researchers included a variable that was the ratio of online credits earned to online courses taken.

The final set of control measures, “school characteristics,” included a nominal indicator of school control (e.g., public, Catholic, or other private). The second variable in this set was a classification of urbanicity (e.g., urban, rural, town, or suburb). Information from the base-year administrator questionnaire pertaining to the degree to which administrators felt dropout was a problem at their school, and an indicator of whether or not their school provided a dropout prevention program was also included in the study. A measure of the percentage of students receiving special education services was also included. Last, a measure of the type of school students attended was included. This variable took on several values from regular noncharter, charter school, special program or interest school, vocational or technical school, and alternative school. All three sets of control measures along with the outcome and predictor variables are listed in Table 1 along with their descriptive information broken out by IEP status.

Analytic Approach

The authors of the current study employed a series of linear probability models to estimate the associations between online coursetaking among SWDs and the high school completion outcomes.¹ The researchers began with a baseline linear probability model that did not include any of the previously described sets of statistical controls in order to ascertain the “unconditional” association between online course participation and the outcome variables. This naive baseline model could be expressed in the following way:

$Y_{i s} = SWD β {+θOL}_{i} + ψSWD \times {OL}_{i} + ∊_{i s},$
1

where on the left side of Equation 1, Y represents a placeholder for either high school dropout or high school graduation for student i in school s. On the other side of the equation, SWD represents a binary indicator of IEP status, OL is the number of online courses a student took in high school, and SWD × OL is the interaction of these two. The coefficient associated with this interaction term, ψ, was the primary parameter of interest. The error term, ∊, represented all other factors explaining the association between online coursetaking among SWDs and the dependent variables. Importantly, to account for correlations among students nested within the same school, this error term was clustered at the school level.

The study followed this baseline with four additional linear probability models. The baseline model was improved upon by first including the set of “demographic characteristics” (Model 2). Model 3 added the set of controls previously referred to as academic characteristics. Model 4 included all three sets of statistical controls listed in Table 1: student demographics, academic characteristics, and “school characteristics.” This final model could be represented as:

$Y_{i s} = SWD β {+ θOL}_{i s} + ψSWD \times {OL}_{i s} + {δSD}_{i} + ς {AC}_{i} + ϕ {SF}_{i s} {+ ∊}_{i s},$
2

where the outcome variable, Y, is once again high school dropout or graduation. On the right side of Equation 2, SD represents student demographics, AC represents academic characteristics, and SF represents school factors. Again, the error term, ∊, was clustered at the school level to account for correlations among student observations nested within the same school.

School Fixed Effects

To reduce the potential threat of omitted variable bias, the researchers improved upon Model 4 by including a school fixed effects term. In so doing, the study was able to control for all observed and unobserved sources of heterogeneity related to the high schools students attended. This was important to do since the current analysis relied on observational data and while great efforts were made to control statistically for school-level factors theoretically proximal to the relationship between the outcome variables and online coursetaking (e.g., urbanicity), it is likely that other school factors—factors not observed by the researchers—could have impacted this relationship. For example, it is likely the case that schools in the sample varied in terms of the resources they had and allocated to dropout prevention. Also, some schools may have been headed by school leaders particularly concerned with the issue of dropout. In terms of online coursetaking, some schools in the sample may have had more robust and developed online courses and/or more talented online teachers relative to other schools. All of these factors could have had a plausible impact on the estimates, yet these factors were not and often cannot be observed. School fixed effects reduce the threat of these omissions by using each school as its own control, thereby focusing analyses on within, rather than between school variation. This school fixed effects model could be expressed as follows:

$Y_{i s} = {SWD}_{i s} β {+ θOL}_{i s} + ψSWD \times {OL}_{i s} + {δSD}_{i} + ς {AC}_{i} {+ γ}_{s} {+ ∊}_{i s},$
3

where Equation 3 is identical to Equation 2 except for the term, γ, which represents school fixed effects.

Results

Table 3 presents the estimated changes in the probability of experiencing a high school dropout episode associated with the main effects and interaction of online coursetaking and IEP status. Coefficients produced by linear probability models can be interpreted as marginal changes in the probability of the outcome for a one unit increase in the predictor (Hellevik, 2009). For example, a coefficient of .03 can be interpreted as increase in the probability of the outcome of 3%. For parsimony and clarity, the researchers have excluded the coefficients associated with the control variables and presented the results of the main effects of online coursetaking and having a disability as well as the interaction between the two.

Table 3.
Linear Probabilities of High School Dropout.

Variables Model 1 Model 2 Model 3 Model 4 Model 5

Courses online .01 (.01) .01 (.01) .00 (.01) .00 (.01) .00 (.01)

IEP .07* (.01) .05* (.01) .02 (.01) .02 (.01) .01 (.01)

IEP × Courses online −.00 (.02) −.00 (.02) .00 (.02) −.00 (.02) .01 (.03)

Controls

Student demographics No Yes Yes Yes Yes

Academic characteristics No No Yes Yes Yes

School factors No No No Yes Yes

School fixed effects No No No No Yes

Observations 7,250 7,250 7,250 7,250 7,250

Note. Standard errors are given in parentheses. IEP = individualized education plan.

p < .05. p < .01. p < .001.

High School Dropout

The first column of Table 3 illustrates that online coursework was not predictive of changes in the probability of experiencing a high school dropout episode. In contrast, the probability of experiencing a high school dropout episode was roughly 7% higher for SWDs than it was for students not receiving special education services. Of greater interest, however, was the coefficient associated with the interaction between online coursework and IEP status. Looking again at Table 3, it is clear that this interaction was not statistically significant, meaning that SWDs who enrolled in online courses were not more or less likely to experience a dropout episode.

The second column in Table 3 conditioned the association of online coursetaking among SWDs and the completion outcomes on the set of student demographic controls. Looking at the results, the main effect of online coursetaking again was not associated with changes in the probability of experiencing a dropout episode among students in the sample. With that said, even after controlling for student demographics, SWDs were still more likely to experience a dropout episode relative to their nondisabled peers, though the size of the coefficient decreased slightly. Importantly, SWDs in the sample who took online courses were not more or less likely to experience a dropout episode in high school.

The third column in Table 3 presents the results of the linear probability models accounting for both student demographics and academic characteristics. Results show that online coursetaking was still unassociated with experiencing a dropout episode. Yet the same was now true for the main effect of IEP status; after controlling for student demographics as well as academic characteristics, SWDs in the sample were no longer associated with increased or decreased probabilities of experiencing a dropout episode in high school. The interaction between disability status and online coursetaking was also not statistically different from 0.

Model 4 included important school factors in addition to student demographics and academic characteristics. There was no change from Models 3 and 4: After controlling for student demographics, academic characteristics, and school factors, the main effects of online coursetaking and IEP status were not statistically associated with changes in the probability of experiencing a high school dropout episode and the same was true for the interaction of these two factors. Column 5 presents the results of the linear probability model with school fixed effects and, as before, after accounting for all observed and unobserved school factors, along with student demographic and academic information, SWDs who chose to take online courses were not more likely or unlikely to experience a dropout episode.

In summary, results indicated that online coursetaking was not associated with increases or decreases in the probability of experiencing a high school dropout episode for SWDs in the sample. This was true regardless of model specification.

High School Graduation

Table 4 presents the changes in the linear probability of high school graduation associated with the main effects of online coursetaking, IEP status, and their interaction. Column 1 contains the results from the baseline, unconditional model. Results contained in this column indicated that online coursetaking was predictive of decreased probabilities of high school graduation. It was also seen that—as extant literature would suggest—SWDs had probabilities of high school graduation that were 9% less than students not receiving special education services. Interestingly though, SWDs who took online courses were 3% more likely to graduate from high school relative to similar students who did not take online courses.

Table 4.
Linear Probabilities of High School Completion.

Variables Model 1 Model 2 Model 3 Model 4 Model 5

Courses online −.02* (.01) −.02 (.01) −.01 (.01) −.01 (.01) −.01 (.01)

IEP −.09* (.01) −.08* (.01) −.03 (.01) −.04 (.01) −.04 (.02)

IEP × Courses Online .03* (.01) .03* (.01) .03 (.01) .03 (.01) .07* (.02)

Controls

Student demographics No Yes Yes Yes Yes

Academic characteristics No No Yes Yes Yes

School factors No No No Yes Yes

School fixed effects No No No No Yes

Observations 7,250 7,250 7,250 7,250 7,250

Note. Standard errors are given in parentheses. IEP = individualized education plan.

As the results in Column 2 show, after controlling for gender, race, and SES, there was little variation in the estimates produced by the baseline model. Online coursetaking was associated with a 2% decrease in the probability of high school graduation and SWDs remained 8% less likely to graduate from high school relative to their peers in the general population. SWDs who completed online courses remained approximately 3% more likely to graduate high school. Moving to Columns 3 and 4, after holding constant the set of student demographic controls, their academic characteristics, and school factors, online coursework was no longer statistically associated with reduced probabilities of high school completion. In Model 3, however, SWDs exhibited probabilities of high school graduation that were approximately 3% less than students not receiving special education services. This association grew to 4% in Model 4, when school factors were accounted for. In both Models 3 and 4, results show that SWDs who completed online courses exhibited probabilities of high school graduation that were 3% higher than SWDs who did not complete online courses.

The fifth column in Table 4 contains the estimates produced by the linear probability models with school fixed effects. Again, online coursetaking was not significantly predictive of changes in the probability of high school graduation. Yet, even after accounting for observed and unobserved school factors and holding constant student demographic and academic information, SWDs exhibited probabilities of high school completion that were 4% less than students not receiving special education services. However, the interaction between online coursetaking and IEP status was more telling and indicated that SWDs who completed online courses had probabilities of high school graduation that were 7% greater than SWDs who did not take online courses. This association was much greater than the models without school fixed effects, which suggests the presence of substantial between-school variation in the data, variation that appears to have biased the previous estimates downward.

In summary, online coursetaking was not, in and of itself, associated with changes in the probability of high school graduation after accounting for academic and school characteristics. In contrast, having an IEP in high school was associated with significant decreases in the probability of graduation even after controlling for demographics, academic characteristics, and all observed and unobserved school factors. Yet the more salient finding is that the interaction of IEP status and online coursetaking was significantly associated with increased probabilities of high school graduation, regardless of model specification.

Discussion

The purpose of this study was to explore the extent to which online coursetaking in high school among SWDs was associated with high school dropout and/or graduation. The study began by noting that despite improvements in the rates of high school graduation and dropout among students not receiving special education services, SWDs continue to fare much worse on these outcomes. Of the numerous interventions seeking to improve graduation and dropout rates, just a few of these interventions have focused on SWDs. A synthesis of this limited body of literature suggests that more personalized learning and increased flexibility may help to close the persistent gaps between SWDs and students not receiving special education services insofar as high school graduation and dropout are concerned (Wilkins & Huckabee, 2014).

The current study argued that despite the many critical findings uncovered by previous empirical studies, learning online holds great potential for decreasing the rate of dropout and increasing the rate of high school graduation among SWDs. This argument was grounded largely on the work of Wilkins and Huckabee (2014) and Dynarski et al. (2008) whose syntheses of the literature suggested that dropout could be attenuated through personalized instruction and increased flexibility, as well as the framework offered by Cavanaugh et al. (2013) and Repetto et al. (2010) who, speaking specifically about SWDs, posited that online learning can reduce the incidence of dropout among SWDs so long as online coursework provides these students with control, effective curriculum, climate, caring community, and connection—the five Cs of Student Engagement. This study also draws heavily from the work and advocacy of organizations like the iNACOL and the COLSD.

Results of the current study found that SWDs who enrolled in online courses throughout high school were not more or less likely to experience a dropout episode in high school relative to their nononline coursetaking peers. These results were robust to model specification. Even after accounting for student demographics, academic background, and all observed at unobserved school factors, SWDs who took online courses exhibited probabilities of high school dropout that were not significantly different than 0.

The current study also confirmed that SWDs were significantly less likely to graduate high school relative to their nondisabled peers. This was a consistent finding across all model specifications. Yet, for SWDs who enrolled in online courses, the opposite was true: SWDs were roughly 7 percentage points more likely to graduate high school after including controls for student demographics, academic background, and all observed and unobserved school factors. In other words, even after controlling for a wide range of students demographic, academic, and school characteristics, SWDs who took online courses in high school exhibited increased probabilities of high school completion.

The current study improves upon what is known about online learning among SWDs by using rich and nationally representative data and a sound empirical strategy. And while this study was not able to provide evidence of causal relationships, the associations uncovered represent a starting point for coming to understand more about an important yet underresearched topic.

There are a number of conclusions to be drawn from the findings presented here. First, even though the current study did not find statistically significant associations between online coursetaking and experiencing a high school dropout episode, results did reveal that SWDs who enrolled in online coursework were, on average, more likely to graduate high school relative to their peers who did not enroll in online coursework. A fair interpretation of these findings is that, among other things, online coursework may provide SWDs with flexibility sufficient to increase their probabilities of school completion. Hence, the findings here appear to lend support for the five Cs framework advanced by Repetto et al. (2010) and Cavanaugh et al. (2013) and support prior work of Basham and Stahl (2014) and Woodworth et al. (2015) who found positive educational outcomes for SWDs in online mathematics courses. Future qualitative analysis should uncover the specific forms that personalization or flexibility may have taken for SWDs. Plausible examples may be that SWDs at greater risk of not completing high school on time or at all benefited from the ability to complete a percentage of his or her course work from home, in a more adapted and comfortable environment. Also, it could be the case that SWDs who struggled interpersonally in traditional classroom settings were better able to navigate social relationships with their teachers or peers online, leading to improved educational outcomes. Rather than frustrate, intimidate, and, consequently, delay the students as traditional classroom environments might, online courses may have provided SWDs with safe climates, greater control, and increased flexibility with regard to how and when they learned.

A second conclusion derived from the current study is that online coursetaking itself was not associated with changes in the probability of high school graduation or dropout after accounting for a rich set of covariates. Rather than being inherently harmful to student success, the degree to which online learning can aid harm completion appears to be moderated by disability status. Of course, there is tremendous variation between schools in the quality of online instruction that students receive. Some students complete online courses taught by veteran online teachers, while others are offered just a limited number of classes, which can be taught by novice and inexperienced teachers. Furthermore, schools vary in the resources they provide their online teachers and students. Some schools, for instance, restrict enrollment in online courses based on academic profiles of their students, while other schools may not have such a mechanism in place while potentially exposing students to suboptimal learning environments.

Further qualitative studies could be conducted to identify the specific mechanisms underlying the broad trends the researchers identified. For example, numerous studies suggest that there is a strong, positive correlation between success and persistence in online courses and students’ internal locus of control. Locus of control refers to students’ belief of the attribution their actions have on an outcome and their control of subsequent events (Lee, Choi, & Kim, 2013). Phirangee and Malec (2017) found that students in online learning communities are more likely to experience feelings of isolation and marginalization when a student’s dominant identity differs from the group. This can negatively impact a student’s learning experience and inhibit a student’s ability to create shared meaning and understanding (Phirangee & Malec, 2017). Furthermore, Greer, Crutchfield, and Woods (2013) posited that the effectiveness of online courses is contingent upon the suitability of its instructional design for the cognitive process of the intended students. Current online course designs are likely to be suitable for students not receiving special education services, but not necessarily SWDs (Greer, Crutchfield, & Woods, 2013).

Future studies might also seek to replicate the analyses performed here after disaggregating by disability category. While the current study offers a glimpse into the associations between online learning among SWDs and school completion, a future study that examines individual disability designations can offer more specific insights such as whether students with physical and mobility disabilities may be actually harmed by online learning, as Crow (2008) suggested.

In light of these conclusions, the current study offers the following policy insights. First, aware that the associations reported here are not causal and that a relatively limited body of literature exists to support the reported conclusions, online learning should be investigated by both researchers and policy makers as a potential mechanism with which to increase high school completion among SWDs. This is a group of students who experience increased rates of dropout and decreased rates of graduation relative to their peers not receiving special education services. Such disparities are evidence, perhaps, of endemic inequities in the educational systems in the nation and as interested parties work to build more just and equitable systems, effective online learning programs should be factored in those discussions. Of course, any and all online programs targeting SWDs must be designed with accessibility in mind. It is often the case that school systems design online programs with the general student population at the forefront of their attention only to later amend or adapt to those programs for SWDs and often when it is too late (Burdette et al., 2013; Roberts et al., 2011). This must change.

A second policy insight to be derived from the current analysis is that schools vary in the development, quality, and effectiveness of their online courses and programs. This variety is likely a determinant of student success since schools with robust online programs and skilled online teachers would be theoretically better positioned to provide quality online coursework for SWDs. Perhaps a logical policy initiative, therefore, would it be to establish or to expand collaboratives or partnerships across schools or districts with the purpose of aiding more resource-scarce institutions with tools, knowledge, and human resources necessary to build engaging and effective online programs. Such partnerships should consider the five Cs framework (Cavanaugh, Repetto, Wayer, & Spitler, 2013; Repetto, Cavanaugh, Wayer, & Liu, 2010) as a foundation as they discuss teaching and learning in online courses.

This report is limited in a number of important ways. First, while there are great benefits with working with large, nationally representative data, the lack of nuanced information regarding students’ online courses along with detailed information related to their experiences and behaviors in these courses restricts the generalizability of the findings presented here. Given the wide variety of technically enhanced coursework available to students today, and without clear information pertaining to the online courses students in the analytic sample completed, the researchers would caution against making categorical claims of support for or against “online learning.” A second limitation of the current study pertains to the treatment of student disability. Specifically, the data set used in this study relied on a broad disability designation, namely, whether a student in the sample had a formalized IEP in the ninth grade. The exact disability and the severity of disability were unknown, yet it is likely that the associations reported here would vary by both disability and severity. Hence, the current authors strongly encourage future research to address these limitations. A third limitation of the current study is the lack of control for teacher quality and course design, two factors that previous research has shown to influence students perceptions of and performance within their online courses (Shea, Sau Li, & Pickett, 2006).

The number of students enrolling in online courses in high school has grown steadily in recent years and forecasts project continued growth. It is inevitably the case that SWDs a community of students who comprise nearly 15% of all students will enroll in these courses along with their peers (McFarland et al., 2017). By offering increased flexibility, personalized instruction, and greater control, online learning holds great potential benefit for these students. If the two most visible measures of student success are high school completion and the avoidance of dropout, it must be acknowledged that SWDs fair much worse on these metrics relative to students without disabilities. However, as this study has sought to examine, online learning may be one vehicle with which to help students press on toward their educational goals. As the former Assistant Secretary of the U.S. Department of Education’s Office of Special Education Programs was quoted as saying, “for most of us, technology makes things easier. For a person with disabilities, it makes things possible” (Qahmash, 2018, p. 1).

	All Students	No IEP	IEP
Outcome variables
Ever dropout	0.10	0.29	0.08	0.27	0.17	0.37
High school completer	0.92	0.28	0.93	0.25	0.84	0.37
Student demographics
Female	0.49	0.50	0.52	0.50	0.36	0.48
White	0.57	0.49	0.57	0.49	0.56	0.50
Black	0.08	0.28	0.07	0.26	0.13	0.34
Asian	0.08	0.27	0.09	0.28	0.04	0.19
Hispanic	0.17	0.37	0.17	0.37	0.17	0.38
Other	0.10	0.30	0.10	0.30	0.10	0.30
Socioeconomic status	0.08	0.80	0.14	0.79	−0.20	0.78
Academic characteristics
School engagement	−0.32	2.03	−0.21	1.91	−0.78	2.43
School belonging	0.02	1.09	0.07	1.06	−0.19	1.17
High school GPA	2.78	0.80	2.87	0.77	2.39	0.79
Ratio of online credits to online courses	0.02	0.11	0.02	0.11	0.02	0.11
Friends who have dropped out	0.29	0.63	0.27	0.60	0.41	0.71
School characteristics
Public school	0.81	0.39	0.78	0.41	0.93	0.25
Catholic school	0.11	0.31	0.13	0.33	0.04	0.20
Other private school	0.08	0.27	0.09	0.29	0.03	0.17
City	0.29	0.45	0.30	0.46	0.23	0.42
Suburb	0.33	0.47	0.33	0.47	0.33	0.47
Town	0.13	0.34	0.14	0.34	0.13	0.34
Rural	0.25	0.43	0.23	0.42	0.31	0.46
Not a problem	0.33	0.47	0.36	0.48	0.21	0.41
Minor problem	0.43	0.50	0.42	0.49	0.48	0.50
Moderate problem	0.17	0.38	0.16	0.37	0.21	0.41
Serious problem	0.06	0.24	0.06	0.23	0.09	0.29
Dropout prevention program	0.39	0.49	0.38	0.49	0.46	0.50
Percent recieving services	11.36	8.02	10.77	7.76	13.85	8.57
Regular noncharter	0.95	0.22	0.95	0.21	0.94	0.24
Charter school	0.03	0.16	0.02	0.16	0.03	0.17
Special program or magnet	0.02	0.14	0.02	0.14	0.02	0.14
Vocational or tech school	0.00	0.06	0.00	0.04	0.01	0.09
Alternative school	0.00	0.03	0.00	0.02	0.00	0.06
Observations	7,250	5,870	1,380

Variables	1–2 Courses	3–5 Courses	6–10 Courses	>10 Courses
Outcome variables
Ever dropout	0.08	0.28	0.17	0.38	0.16	0.37	0.23	0.43
High school completer	0.95	0.23	0.88	0.33	0.93	0.26	0.81	0.40
Student demographics
Female	0.51	0.50	0.51	0.50	0.54	0.50	0.71	0.46
White	0.52	0.50	0.50	0.50	0.54	0.50	0.60	0.50
Black	0.11	0.31	0.09	0.29	0.18	0.39	0.09	0.28
Asian	0.09	0.29	0.08	0.27	0.08	0.28	0.00	0.00
Hispanic	0.16	0.37	0.18	0.38	0.13	0.34	0.20	0.41
Other	0.12	0.32	0.15	0.36	0.07	0.25	0.11	0.32
Socioeconomic status	0.11	0.83	0.04	0.74	0.10	0.78	−0.04	0.66
Academic characteristics
School engagement	−0.78	2.61	−0.82	2.43	−0.75	2.45	−0.09	1.64
School belonging	0.10	0.95	−0.11	1.06	0.07	1.17	−0.03	1.20
High school GPA	2.72	0.84	2.48	0.89	2.65	0.94	2.47	0.76
Ratio of online credits to online courses	0.53	0.26	0.50	0.24	0.50	0.19	0.41	0.13
Friends who have dropped out	0.26	0.58	0.42	0.68	0.57	0.79	0.63	0.93
School characteristics
Public school	0.85	0.36	0.88	0.33	0.83	0.38	0.74	0.44
Catholic school	0.05	0.22	0.02	0.13	0.05	0.21	0.06	0.24
Other private school	0.10	0.30	0.11	0.31	0.13	0.34	0.20	0.41
City	0.30	0.46	0.28	0.45	0.30	0.46	0.54	0.51
Suburb	0.39	0.49	0.44	0.50	0.35	0.48	0.09	0.28
Town	0.07	0.25	0.06	0.24	0.06	0.25	0.09	0.28
Rural	0.24	0.43	0.21	0.41	0.29	0.46	0.29	0.46
Not a problem	0.30	0.46	0.29	0.45	0.32	0.47	0.36	0.49
Minor problem	0.46	0.50	0.46	0.50	0.53	0.50	0.42	0.50
Moderate problem	0.21	0.41	0.18	0.39	0.11	0.32	0.12	0.33
Serious problem	0.03	0.17	0.07	0.25	0.04	0.19	0.09	0.29
Dropout prevention program	0.44	0.50	0.36	0.48	0.39	0.49	0.36	0.49
Percent receiving services	11.75	7.66	12.41	7.18	11.77	9.16	11.50	8.56
Regular noncharter	0.95	0.22	0.91	0.29	0.91	0.29	0.94	0.24
Charter school	0.02	0.14	0.02	0.16	0.05	0.23	0.03	0.17
Special program or magnet	0.02	0.15	0.02	0.16	0.02	0.13	0.00	0.00
Vocational or tech school	0.00	0.06	0.00	0.00	0.00	0.00	0.00	0.00
Alternative school	0.01	0.08	0.04	0.20	0.02	0.13	0.03	0.17
Observations	607		176		63		35

Variables	Model 1	Model 2	Model 3	Model 4	Model 5
Courses online	.01 (.01)	.01 (.01)	.00 (.01)	.00 (.01)	.00 (.01)
IEP	.07*** (.01)	.05*** (.01)	.02 (.01)	.02 (.01)	.01 (.01)
IEP × Courses online	−.00 (.02)	−.00 (.02)	.00 (.02)	−.00 (.02)	.01 (.03)
Controls
Student demographics	No	Yes	Yes	Yes	Yes
Academic characteristics	No	No	Yes	Yes	Yes
School factors	No	No	No	Yes	Yes
School fixed effects	No	No	No	No	Yes
Observations	7,250	7,250	7,250	7,250	7,250

Variables	Model 1	Model 2	Model 3	Model 4	Model 5
Courses online	−.02*** (.01)	−.02** (.01)	−.01 (.01)	−.01 (.01)	−.01 (.01)
IEP	−.09*** (.01)	−.08*** (.01)	−.03** (.01)	−.04** (.01)	−.04** (.02)
IEP × Courses Online	.03*** (.01)	.03*** (.01)	.03** (.01)	.03** (.01)	.07*** (.02)
Controls
Student demographics	No	Yes	Yes	Yes	Yes
Academic characteristics	No	No	Yes	Yes	Yes
School factors	No	No	No	Yes	Yes
School fixed effects	No	No	No	No	Yes
Observations	7,250	7,250	7,250	7,250	7,250

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Cameron Sublett

Note

References

Ahn

McEachin

(2017). Student enrollment patterns and achievement in Ohio’s online charter schools. Educational Researcher, 46, 44–57.

Basham

J. D.

Stahl

(2014, 4). Supporting outcomes for all students in personalized learning environments. Presented at the Annual Meeting of the American Educational Research Association, Philadelphia, PA.

Barbour

M. K.

Reeves

T. C.

(2009). The reality of virtual schools: A review of the literature. Computers & Education, 52, 402–416. doi:10.1016/j.compedu.2008.09.009

Belfield

C. R.

Levin

H. M.

(2007). The price we pay: Economic and social consequences of inadequate education. Washington, DC: Brookings Institution Press.

Belz

J. A.

Müller-Hartmann

(2003). Teachers as intercultural learners: Negotiating German-American telecollaboration along the institutional fault line. Modern Language Journal, 87, 71–89.

Burdette

Greer

P. L.

Woods

Kari

(2013). K-12 Online learning and students with disabilities: Perspectives from state special education directors. Journal of Asynchronous Learning Networks, 17, 65–72.

Carr

(2000). As distance education comes of age, the challenge is keeping the students. Chronicle of Higher Education, 46, A39–A41. Retrieved from http://chronicle.com/section/Home/5/?eio=34361

Carsen

(2016, 12 19). Alabama admits its high school graduation rate was inflated. National Public Radio Education. Retrieved from http://www.npr.org/sections/ed/2016/12/19/505729524/alabama-admits-its-high-school-graduation-rate-was-inflated

Cavanaugh

Gillan

K. J.

Kromrey

Hess

Blomeyer

(2004). The effects of distance education on K–12 student outcomes: A meta-analysis. Naperville, IL: Learning Point Associates.

10.

Cavanaugh

Repetto

Wayer

Spitler

(2013). Online learning for students with disabilities: A framework for success. Journal of Special Education Technology, 28, 1–8.

11.

Christensen

Horn

Johnson

(2008). Disrupting class: How disruptive innovation will change the way the world learns. New York, NY: McGraw-Hill.

12.

Crow

(2008). Four types of disabilities: Their impact on online learning. TechTrends, 52, 51–55.

13.

De la Varre

Irvin

M. J.

Jordan

A. W.

Hannum

W. H.

Farmer

T. W.

(2014). Reasons for student dropout in an online course in a rural K–12 setting. Distance Education, 35, 324–344. doi:10.1080/01587919.2015.955259

14.

Dupin-Bryant

P. A.

(2004). Pre-entry variables related to retention in online distance education. American Journal of Distance Education, 18, 199–206. doi:10.1207/s15389286ajde1804_2

15.

Dynarski

Clarke

Cobb

Finn

Rumberger

Smink

. (2008). Dropout prevention. IES practice guide. NCEE 2008-4025. Washington, DC: National Center for Education Evaluation and Regional Assistance. Retrieved from http://files.eric.ed.gov/fulltext/ED502502.pdf

16.

Gabriel

(2011, 4 5). Rise in online classes flares debate about quality. The New York Times. Retrieved from https://www.nytimes.com/2011/04/06/education/06online.html

17.

Gemin

Pape

Vashaw

Watson

(2015). Keeping pace with K-12 digital learning: An annual review of policy and practice. Evergreen Education Group. Retrieved from http://www.kpk12.com/wp-content/uploads/Evergreen_KeepingPace_2015.pdf

18.

Gill

Walsh

Smither Wulsin

Matulewicz

Severn

Grau

… Kerwin

. (2015). Inside online charter schools. Mathematica Policy Research. Retrieved from https://www.mathematica-mpr.com/our-publications-and-findings/publications/inside-online-charter-schools

19.

Greer

D. L.

Crutchfield

S. A.

Woods

K. L.

(2013). Cognitive theory of multimedia learning, instructional design principles, and students with learning disabilities in computer-based and online learning environments. Journal of Education, 193, 41–50.

20.

Gross

Tuchman

Patrick

(2018). A national landscape scan of personalized learning in K-12 education in the United States. Vienna, VA: iNACOL.

21.

Hashey

A. I.

Stahl

(2014). Making online learning accessible for students with disabilities. Teaching Exceptional Children, 46, 70–78.

22.

Heckman

J. J.

Lafontaine

P. A.

(2010). The American high school graduation rate: Trends and levels. The Review of Economics and Statistics, 92, 244–262.

23.

Heissel

(2016). The relative benefits of live versus online delivery: Evidence from virtual algebra I in North Carolina. Economics of Education Review, 53, 99–115.

24.

Hellevik

(2009). Linear versus logistic regression when the dependent variable is a dichotomy. Quality & Quantity, 43, 59–74.

25.

Ingles

S. J.

Prat

D. J.

Herget

D. R.

Bryan

Fritch

L. B.

Ottem

… Wilson

(2015). High school longitudinal study of 2009 (HSLS:09) 2013 update and high school transcript data file documentation (No. NCES 2015-036). National Center for Education Statistics, Institute of Education Sciences, U.S. Department of Education. Retrieved from https://nces.ed.gov/pubs2015/2015036.pdf

26.

International Association for K-12 Learning. (2013). The online learning definitions project. Retrieved from https://www.inacol.org/resource/the-online-learning-definitions-project/

27.

Kamenetz

Turner

(2016, 10 17). The high school graduation rate reaches a record high—Again. National Public Radio Education. Retrieved from http://www.npr.org/sections/ed/2016/10/17/498246451/the-high-school-graduation-reaches-a-record-high-again

28.

Lee

Choi

Kim

(2013). Discriminating factors between completers of and dropouts from online learning courses. British Journal Of Educational Technology, 44, 328–337.

29.

McFarland

Hussar

de Brey

Snyder

Wang

Wilkinson-Flicker

… Hinz

(2017). The condition of education 2017 (No. NCES 2017-144). Washington, DC: National Center for Education Statistics, U.S. Department of Education. Retrieved from https://nces.ed.gov/pubs2017/2017144.pdf

30.

McFarland

Stark

Cui

(2016). Trends in high school dropout and completion rates in the United States: 2013 (No. NCES 2016-117). Washington, DC: National Center for Education Statistics, U.S. Department of Education. Retrieved from https://nces.ed.gov/pubs2016/2016117rev.pdf

31.

McCleary

(2002). Using multiple imputation for analysis of incomplete data in clinical research. Nursing Research, 51, 339–343.

32.

National Center for Education Statistics. (2017). Public high school 4-year adjusted cohort graduation rate (ACGR), by race/ethnicity and selected demographic characteristics for the United States, the 50 states, and the District of Columbia: School year 2015-16. Retrieved from https://nces.ed.gov/ccd/tables/ACGR_RE_and_characteristics_2015-16.asp

33.

Patrick

Kennedy

Powell

(2013). Mean what you say: Defining and integrating personalized, blended and competency education. International association for K-12 online learning. Retrieved from https://www.inacol.org/resource/mean-what-you-say-defining-and-integrating-personalized-blended-and-competency-education/

34.

Phirangee

Malec

(2017). Othering in online learning: An examination of social presence, identity, and sense of community. Distance Education, 38, 160–172.

35.

Picciano

Seman

(2010). Class connections: High school reform and the role of online learning. Retrieved from https://www.inacol.org/resource/class-connections-high-school-reform-and-the-role-of-online-learning/

36.

Picciano

Seman

Day

(2011). Online learning in Illinois high schools: Has the time come? Retrieved from https://files.eric.ed.gov/fulltext/ED535142.pdf

37.

Qahmash

A. I. M.

(2018). The potentials of using mobile technology in teaching individuals with learning disabilities: A review of special education technology literature. TechTrends, 1–7. doi:10.1007/s11528-018-0298-1

38.

Repetto

Cavanaugh

Wayer

Liu

(2010). Virtual high schools: Improving outcomes for students with disabilities. Quarterly Review of Distance Education, 11, 91–104.

39.

Roberts

J. B.

Crittenden

L. A.

Crittenden

J. C.

(2011). Students with disabilities and online learning: A cross-institutional study of perceived satisfaction with accessibility compliance and services. Internet & Higher Education, 14, 242–250.

40.

Rouse

C. E.

(2007). Quantifying the costs of inadequate education: Consequences of the labor market. In Belfield

C. R.

Levin

H. M.

(Eds.), The price we pay: Economic and Social consequences of inadequate education (pp. 99–124). Washington, DC: Brookings Institution Press.

41.

Royston

(2004). Multiple imputation of missing values. The Stata Journal, 4, 227–241.

42.

Rumberger

R. W.

(2011). Dropping out: Why students drop out of high school and what can be done about it. Cambridge, MA: Harvard University Press.

43.

Shah

(2011). E-Learning access for special needs. Education Week, 31, S2–S4.

44.

Shea

C. S.

Pickett

(2006). A study of teaching presence and student sense of learning community in fully online and web-enhanced college courses. The Internet and Higher Education, 9, 175–190.

45.

Skårbrevik

K. J.

(2005). The quality of special education for students with special needs in ordinary classes. European Journal of Special Needs Education, 20, 387–401. Retrieved from http://10.0.4.56/08856250500268601

46.

Smith

S. J.

Basham

J. D.

(2014). Designing online learning opportunities for students with disabilities. Teaching Exceptional Children, 46, 127–137.

47.

Smith

Clark

Blomeyer

(2005). A synthesis of new research on K-12 online learning (p. 96). Naperville, IL: Learning Point Associates.

48.

Snart

(2017). Hybrid learning at the community college. New Directions for Teaching & Learning, 2017, 59–67. Retrieved from http://10.0.3.234/tl.20227

49.

Sparks

S. D.

(2016). Many students “stop out” of high school, studies show. Education Week, 36, 7–7.

50.

Stark

Noel

A. M.

(2015). Trends in high school dropout and completion rates in the United States: 1972–2012 (No. NCES 2015-015). Washington, DC: National Center for Education Statistics, U.S. Department of Education. Retrieved from https://nces.ed.gov/pubs2015/2015015.pdf

51.

Sublett

Rumberger

(2017). Accounting for delayed high school graduates in California. California Dropout Research Project (Policy Report #22). Retrieved from http://cdrpsb.org/pubs_reports.htm

52.

Sum

Khatiwada

McLaughlin

(2009). The consequences of dropping out of high school: Joblessness and jailing for high school dropouts and the high cost for taxpayers. Center for Labor Market Studies. Retrieved from https://repository.library.northeastern.edu/downloads/neu:376324?datastream_id=content

53.

Tallent-Runnels

M. K.

Thomas

J. A.

Lan

W. Y.

Cooper

Ahern

T. C.

Shaw

S. M.

Liu

(2006). Teaching courses online: A review of the research. Review of Educational Research, 76, 93–135. doi:10.3102/00346543076001093

54.

The Center on Online Learning and Students with Disabilities. (2016). Equity matters 2016: Digital & online learning for students with disabilities. Retrieved from http://www.centerononlinelearning.res.ku.edu/equity-matters-2016-digital-and-online-learning-for-students-with-disabilities/

55.

The Foundation for Blended and Online Learning and Evergreen Education Group. (2007). Why do students choose blended and online schools: The end of average requires personalized learning environments (no. The Foundation for Blended and Online Learning). Retrieved from http://www.blendedandonlinelearning.org/wp-content/uploads/2017/01/FBOL-Why-Students-Choose-Blended-Online-Schools-Jan-2017.pdf

56.

Tindle

East

Mellard

(2017). Online learning for students with disabilities: Considerations for SEA policies and procedures. Lawrence, KS: Center on Online Learning and Students with Disabilities.

57.

Watson

J. F.

Kalmon

(2005). Keeping pace with K–12 online learning: A review of state-level policy and practice. Naperville, IL: Learning Point Associates. Retrieved August 4, 2011, from http://www.learningpt.org/pdfs/tech/Keeping_Pace2.pdf

58.

Watson

Murin

Vashaw

Gemin

Rapp

(2011). Keeping pace with K-12 online learning: An annual review of policy and practice. Evergreen, CO: Evergreen Education Group. Retrieved from https://www.evergreenedgroup.com/keeping-pace-reports/

59.

Wilkins

Huckabee

(2014). A literature map of dropout prevention interventions for students with disabilities. Prevention Center for Students with Disabilities. Retrieved from https://eric.ed.gov/?id=ED571791

60.

Woodworth

J. L.

Raymond

M. E.

Chirbas

Gonzalez

Negassi

Snow

Van Donge

(2015). Online charter school study 2015. Retrieved from: https://virtualschooling.wordpress.com/2015/12/10/special-issue-of-journal-of-online-learning-research/

61.

Wojciechowski

Palmer

L. B.

(2005). Individual student characteristics: Can any be predictors of success in online classes? Online Journal of Distance Learning Administration, 8. Retrieved from http://www.westga.edu/∼distance/ojdla/