Research on optimal class size for online education
Stephen Fox, PhD
University of Hawaii Maui College
“Participation is an important factor in achieving a desirable outcome in higher-level learning.” (Kim, 2013, p. 123)
Online education is increasing at a phenomenal rate, with increases in both online offerings from traditional schools and from entirely online institutions where degrees are completed at distance. Hybrid classes with a mix of online and physical conditions are also popular, including the “flipped classroom” in which material is delivered online with in-person classes serving as time for activities and elaboration. The latter have been shown to be especially effective, possibly because of the interactive and supportive nature of the reinforcing activities (Lewis & Harrison, 2012). Simultaneously, an initiative has been proposed by President Obama in which schools would be rated on retention and completion rates, highlighting an area of particular risk in online education. Students fail to pass or complete online courses more often than in-person classes, a trend that may only be countered by specific intervention and mentoring (Gaskell, 2006; Waters, 2013).
Three interrelated issues predominate concerns in research on class size: retention, interaction, and quality. Students want and need quality delivery to have a successful matriculation and subsequent outcomes. In terms of retention, without other factors intervening, previous student success predicts retention (Cochran, Campbell, & Baker, 2013). Hixenbaugh, Dewart, & Towell, (2012) found that student physical health, well-being, and social support were primary factors in retention. If completion is to be the measure of success and funding, the simple solution is to raise entry standards, which will disenfranchise a massive portion of potential students. The alternative is to find methods to improve student success even amongst those with previous low performance and insufficient social support. Given the confluence of increasing demand for online classes and placement of accountability for completion at the institutional level, serious analysis is warranted. Triangulating with retention, quality of instruction and interaction are obvious areas of focus.
As Kim (2013) states, participation is a key element in outcomes, but participation may vary from simply completing assignments to synchronous interaction between students and instructors. In Boettcher’s (2013) Best Practices compendium, practice number 5 suggests use of both synchronous and asynchronous activities, which would typically include a combination of posted asynchronous discussions along with some form of synchronous participation in a textual chat or other voice or video situation. Size of class affects both of these, with systems such as Blackboard Collaborate breaking down at over 20 participants, which is about the same range where textual chat becomes unwieldy. Personalized instructor interaction is particularly time consuming, but satisfying to students. Item 5 in Hanover Research Groups (2009) Checklist for Online Interactive Learning (COIL) states that institutions should “Mandate smaller class sizes for online courses to give faculty appropriate time to deliver quality instruction” (p. 6). A surprisingly small number of journal articles address the issue of class size, despite the rapid increase of online learning in recent years and the huge increase in class sizes as Massive Open Online Courses (MOOC’s) are increasingly seen as an economically viable option for lower cost education.
Optimal Class Size (OCS)
Current actual class sizes range from very small (2 to 5) to MOOC’s of hundreds of thousands. Retention, grade, and completion begin to decrease markedly at relatively low numbers. Oestmann and Oestmann (2007) reported an optimal online class size of 20 to 25. In classes of less than 10 or greater than 25, the amount and quality of interaction decreased, and the resulting grade outcomes were reduced. Students were more likely to disengage when enrollment passed 25 (Meredith, pp. 9-10). In her review of available studies, Artz (2011) cites a number of factors that influence outcomes beyond size. For new instructors, a class of 12 appears optimal. In general, sizes of 16 to 19 students appear to have best outcomes. Overall, class size gauged for success appears to be remaining constant or going down across time.
Orleana, Instructors’ Perceptions of Optimal Class Sizes
|Optimal class size||7||80||18.9||9.1|
|OCS if interaction goals achievable||5||50||15.9||6.6|
|Actual class size||4||81||22.8||13.7|
Class size by publication year
|Vrasidas and McIssac||1999||Not too small||Interaction decreases|
|Jones, Ravid, & Rafaeli||2004||Not too big||Too many discussion posts overwhelm|
|Arbaugh & Benbunnan-Finch||2005||25-30|
|Tomei||2006||12||Smaller size generates more detailed messages|
|Rovai||2007||min 8–10, max 20–30||Good interactions|
|Qiu, Hewitt, & Brett||2012||13-15||Smaller groups benefit class discussions|
A primary method of interaction is asynchronous discussion, which brings benefit of engagement and elaboration to increase learning outcomes. Berry’s (2008) review of research suggests that discussions work best in very small groups less than ten, or in classes of 15 divided into smaller groups, though some studies recommend larger class sizes. He cites Fisher, Thompson and Silverberg (2005), who recommended 25 students as the optimal class size for such a discussion. Their two year study found that this group size, which they deem large, produced an optimal number of acceptable messages in discussions. Also cited is Reonieri (2006), whose surveys of faculty and students concluded that 10 – 15 students per class is optimal. Smaller class sizes lacked diversity of views and larger classes became effective only when “divided into the optimally sized groups so that all voices could be heard.” (Berry, 2008, p.1)
Generally, larger class sizes may generate more messages, but not improved or uniformly distributed interactions. Arguello et al. (2006) found that while lager classes generated a high volume of messages, these tended to be posted by a relatively small proportion of the students who dominated discussions (Caspi, Gorsky, & Chajut, 2003; Abuseileek, 2012). Kim (2013) designed a study empirically to test effect of group size on discussion participation and interactivity. In a class totaling 138, students either discussed as a whole class or were divided into groups of 25-30 across the semester. In the smaller groups, both postings and interactivity were consistently much higher, with a particularly greater level of interactive posts in the small group discussions. Kim concludes that his study confirms earlier research “that small grouping itself encourages more interactive participation since the activity of reading in a small group does not remain to be passive and does encompass engagement, thoughts and reflection (Hrastinski, 2009)” (Kim 2013, p. 127).
Social support as a factor in student success online seems to receive only passing mention, despite a general feeling of benefit. Interestingly, anonymity of posting without personal interaction has been observed to be beneficial to certain students (Higgs, 2012). Social support online can come from student or instructor, but can become toxic if not monitored on a fairly constant basis, based on my own experience.
As previously mentioned, both frequency and quality of interaction are highly relevant in student outcomes. An avenue of future research would be to investigate what kinds of interaction, and with whom, lead to better outcomes. My prediction would be that frequent interaction, in a class of around 20 students, that can be monitored on a regular basis by the instructor, will result in sufficient social support delivered along with quality content, leading to optimal retention and completion. Not mentioned is the reduced cost of material infrastructure, given the elimination of physical space needed, which should be counterbalanced against need for more (but probably less expensive) expenditure on technological infrastructure and support. Online instruction involves an inevitable amount of tech support provided by the instructor, which should be included in calculation of instructional time spent, and which could be provided more effectively by IT support personnel, allowing the instructor to specialize in relevant delivery of information and social support.
Abuseileek, A. F. (2012). The effect of computer-assisted cooperative learning methods and group size on the EFL learners’ achievement in communication skills. Computers & Education, 58, 231–239.
Arguello, J., Butler, B. S., Joyce, E., Kraut, R., Ling, K. S., Rose, C., et al. (2006). Talk to me: foundations for successful individual-group interactions in online communities. In Proceedings of the SIGCHI conference on human factors in computing systems, New York, MY.
Artz, J. (2011). Online Courses and Optimal Class Size: A Complex Formula. http://files.eric.ed.gov/fulltext/ED529663.pdf
Berry, G. (2008). Asynchronous Discussions: Best Practices. 24th Annual Conference on Distance Teaching & Learning. http://www.uwex.edu/disted/conference/resource_library/proceedings/08_12701.pdf
Boettcher, J. V. (2013). Ten Best Practices for Teaching Online Quick Guide for New Online faculty. Designing for Learning 2006 – 2013. http://www.designingforlearning.info/services/writing/ecoach/tenbest.html
Caspi,A., Gorsky, P.,&Chajut, E. (2003). The influence of group size on mandatory asynchronous instructional discussion groups. The Internet and Higher Education, 6, 227–240.
Cochran, J. D., Campbell, S. M., Baker, H. M., & Leeds, E. M. (2013). The role of student characteristics in predicting retention in online courses. Research in Higher Education. Advance online publication. doi:10.1007/s11162-013-9305-8
Gaskell , A. (2006). Rethinking access, success and student retention in Open and Distance Learning. Open Learning, 21(2), 95-98. doi: 10.1080/02680510600712997
Glenn, L. M. and Berry, G. R. (2006). Online Best Practice: Interaction Matters. Journal of Business Inquiry.
Hanover Research Group (2009). Best Practices in Online Teaching Strategies. http://www.uwec.edu/AcadAff/resources/edtech/upload/Best-Practices-in-Online-Teaching-Strategies-Membership.pdf
Higgs, A. (2012). E-learning, ethics and ‘non-traditional’ students: Space to think aloud. Ethics and Social Welfare, 6(4), 386-402. doi:10.1080/17496535.2012.654496
Hixenbaugh, P., Dewart, H., & Towell, T. (2012). What enables students to succeed? An investigation of socio-demographic, health and student experience variables. Psychodynamic Practice: Individuals, Groups and Organisations, 18(3), 285-301. doi:10.1080/14753634.2012.695887
Hrastinski, S. (2009). A theory of online learning as online participation. Computers & Education, 52, 78–82.
Kim, J. (2013). Influence of group size on students’ participation in online discussion forums. Computers & Education, 62, 123-129. doi:10.1016/j.compedu.2012.10.025
Lee, K. C., Chung, N., & Lee, S. (2011). Exploring the influence of personal schema on trust transfer and switching costs in brick-and-click bookstores. Information & Management, 48(8), 364-370. doi:10.1016/j.im.2011.09.002
Lewis, J. S., & Harrison, M. A. (2012). Online delivery as a course adjunct promotes active learning and student success. Teaching of Psychology, 39(1), 72-76. doi:10.1177/0098628311430641
McCarthy, J. W., Smith, J. L., & DeLuca, D. (2010). Using online discussion boards with large and small groups to enhance learning of assistive technology. Journal of Computing in Higher Education, 22(2), 95-113. doi:10.1007/s12528-010-9031-6
Meredith, B. P. (?). Online Course Class Sizes: A Review of Current Research on the Optimal Size of the Online Classroom. http://wiki.utep.edu/download/attachments/39191673/Online+Course+Class+Sizes_A+Review+of+Current+Research+on+the+Optimal+Size+of+the+Online+Classroom.pdf
Orellana, A. (2006). Instructors’ Perceptions of Optimal Class Sizes. http://www.fischlerschool.nova.edu/resources/uploads/app/72/files/pdf/2007-fall/BPOL-fall07-orellana.pdf
Qiu, M., Hewitt, J., & Brett, C. (2012). Online class size, note reading, note writing and collaborative discourse. International Journal of Computer-Supported Collaborative Learning, 7(3), 423-442. doi:10.1007/s11412-012-9151-2
Waters, J. K. (09/03/13). SJSU MOOC Study Reveals Achievement Gains but Low Retention Rates. Campus Technology. http://campustechnology.com/articles/2013/09/03/sjsu-mooc-study-reveals-achievement-gains-but-low-retention-rates.aspx