Higher education has evolved over recent decades to effectively use technology and move many classes to a full or partial online format. Between 2012 and 2015, online enrollment in the USA has increased 30% (Allen & Seaman,
2015), and as of 2015, 43.1% of undergraduate students have taken at least one online course (National Center for Education Statistics,
2018). With the COVID-19 pandemic, higher education experienced the abrupt pivot to remote teaching. Now, more than ever, it is imperative that we continue to investigate the effects of online instruction on student learning and compare learning gains to in-person instruction.
Online instruction can take many forms. Prior to the advent of the internet, distance education (DE) took the form of correspondence education, where materials were exchanged between student and instructor via mail (Kentnor,
2015). As defined by Gallagher and McCormick (
1999), DE takes place when there is at least a semi-permanent separation between the learner and the instructor. In the 1990s, Internet capabilities allowed for many different forms of DE (Phipps & Merisotis,
2000) and further contributed flexibility to in-person courses (e.g., blended, hybrid, flipped). A blended course uses digital resources to supplement learning that are not meant to replace in-person instruction; whereas a hybrid course implements digital content to replace part of the in-person instructional time (Siegelman,
2020). A flipped classroom can be blended or hybrid, but it may also lack digital materials (see Jensen et al.,
2018). Regardless, a flipped course uses curriculum designed so that students learn basic materials asynchronously before class (most traditional through digital media but can also be through textbook learning) and then practice applying that content in an in-person setting (Hamdan et al.,
2013). A fully online course, on the other hand, is one in which there is a permanent separation of the learner and instructor and all course interactions, whether synchronous or asynchronous, take place in a digital format (Means et al.,
2010). The sheer multitude of variations of an online course design undoubtedly contributes to the mixed findings the education research community reports in comparing online with in-person instruction.
Beyond differences in online versus in-person course delivery, a host of other influential factors such as learner time-on-task, pedagogical approach, or curricular materials (Means et al.,
2010) confound findings across studies. In fact, Zhao et al. (
2005) found effect sizes comparing online to in-person instruction were dependent upon instructor identity, instructor involvement, the outcome measure, student education level, content area, level of instruction, degree of synchronous interactions, and degree of in-person teaching. The purpose and contribution of this study is to control for many of these features by comparing a fully online asynchronous course (i.e., where no synchronous interaction takes place) to a flipped version of an in-person course while maintaining equivalent pedagogical approaches, instructors, time-on-task, and curricular materials. Our guiding research question was, are student academic outcomes equivalent when simply transferring the in-person assignments, activities, and assessments of a flipped classroom directly to a fully online asynchronous format?
Comparisons of Online and In-Person Instruction
Many meta-analyses have been performed on DE methods. However, to determine the effectiveness of online education in comparison to in-person instruction, we only summarize those meta-analyses that included studies that directly compared online classrooms with in-person classrooms and excluded studies comparing online modalities to each other or to hybrid classrooms. As one of the first influential meta-analyses in this category, Shachar and Neumann (
2003) meta-analyzed 86 studies between 1990 and 2002 that directly compared a DE course to an in-person course. In 66% of their studies, DE outperformed in-person instruction models (average effect size of
d = 0.37; Shachar & Neumann,
2003); however, criteria for inclusion as “online” were not articulated and could bias their findings.
The following year, Bernard et al. (
2004) performed a similar meta-analysis comparing DE with “classroom instruction,” except that they expanded the search window to include studies dating back to 1985. In addition, the authors set criteria where their DE classes for study inclusion had to report less than 50% in-person instruction and potential studies had to measure one of the following outcomes: achievement, attitudes, or retention. While these criteria increased transparency beyond the Shachar and Neumann (
2003) meta-analysis, these criteria still lumped hybrid and fully online courses as equivalent DE delivery methods. They also made a further distinction of whether the DE courses included only asynchronous interactions or at least some synchronous interactions. For achievement, Bernard et al. (
2004) found an overall slightly positive effect of DE over in-person instruction; however, the effect sizes and directionality differed by mode with the majority of effect sizes for synchronous DE being negative, while asynchronous effects were positive. For attitudes, Bernard et al. (
2004) found that students in DE courses had significantly poorer attitudes, especially in a synchronous format. And lastly, retention had a slight but significant negative effect favoring in-person instruction but with wide variations in effect sizes between studies (Bernard et al.,
2004).
Similarly, Zhao et al. (
2005) analyzed 51 studies published between 1966 to 2002 including a wide range of DE methodologies all classified as education at a distance, yet they controlled for several moderators in their analysis to quantify their influence on effect sizes including the year of publication, whether the researcher was also the instructor, the instructor’s involvement in the course, level of student (high school or college), and whether or not any instruction occurred face-to-face. Overall, two-thirds of their included studies demonstrated a positive effect favoring DE over in-person instruction (Zhao et al.,
2005). However, when compared with the magnitude of the one-third that had a negative effect, the overall effect size was zero. Zhao et al. (
2005) describe that higher effect sizes were associated with analysis factors (i.e., articles published post-1998), factors of the study participant populations (i.e., high school students were higher than college students), curricular design factors (i.e., greater involvement of the instructor in the DE course, greater proportion of in-person instruction), and potential bias factors (i.e., if the authors was the instructor of the DE course). Again, this analysis is complicated by the wide range of online methodologies included in the analysis.
In 2009, the US Department of Education conducted a thorough review of online learning (Means et al.,
2010), and although they included many forms of DE, their analysis tightly controlled for fully online courses versus blended or hybrid courses. They analyzed 99 studies conducted between 1996 and 2008 that compared in-person instruction to some form of DE. Of the 45 with sufficient data, in-person instruction was compared to fully online instruction in 27 cases and to blended formats in 23 cases (Means et al.,
2010). They concluded that overall, online conditions performed modestly better than in-person, blended courses significantly outperformed in-person courses (g = + 0.35), yet fully online courses had no significant advantage over in-person courses. However, the authors point to the varied conditions of the curricula (time-on-task, pedagogical approaches, etc.) for this modest improvement by delivery method; furthermore, this analysis showed that the more similar the curricula were in the comparison, the smaller the effect size (Means et al.,
2010).
From these meta-analyses, we can see a positive trend for online instruction, with wide variety in effect sizes. However, little, if any, information is known about the comparison between DE and flipped classrooms. This suggests the need for more controlled and rigorous approaches to evaluating DE (Lee,
2017). In this study, we compare two courses that are practically equivalent in terms of design, content, implementation, narrative, assessment, and even instructor affect, utilizing a blended, technology-enhanced flipped classroom as our in-person and a fully online (asynchronous) comparison course.
Problem Statement
Online education is increasingly being incorporated into the undergraduate experience, and the current state of education demands that we offer this delivery method, especially in light of recent events (i.e., a global pandemic that has forced the closure of all institutes of higher education during Spring 2020). However, research has failed to adequately test whether curriculum-equivalent courses in an in-person and asynchronous online modality produce disparate learning outcomes. In a tightly controlled study of nearly equivalent courses, we aimed to investigate whether a baseline model of an asynchronous online course (where equivalent curricula are moved directly online, and instructor/peer interaction was limited to asynchronous communication that required student initiation) leads to equivalent student performance.