Is hy-flex teaching sustainable post-COVID? Insights from an Economics department

The paper presents a model for redesigning undergraduate economics teaching in response to the pandemic uncertainties. The model was formulated and adopted by the Economics Department at Royal Holloway, University of London (RHUL Econ) in the Summer of 2020. The tactical goal of the redesign was to prevent a blow-up of teaching workload in an environment of social distancing and capacity constraints. Its main strategic goal was to gear module content and assessment towards improved employability. The paper discusses the objectives and constraints the department was operating with, and the details of the adopted model. It also scrutinises the short-term sunk costs and the potential long-term benefits from the redesign for both students and staff. It concludes with a review of the scalability of such redesign and with insights into those elements of the model that survived the last 3 years. Those are likely to become a sustainable element of the post-pandemic teaching model across many Economics departments.

Key words: Hy-flex, COVID-19, post-covid education, teaching workload, flipped classroom

I. Introduction

The widespread move to online education and assessment in March 2020 prompted a discussion going beyond the immediate crisis response. Beyond the urgency, the entire education sector faced a more strategic question: how to remodel teaching and assessment for the new normal post-pandemic mode of operation? To address the question, in early April 2020 RHUL Econ researched best-practice online and hybrid pedagogy. The motivation for this research was the need to embed flexibility and resilience in our existing teaching model. The key was to enable a quick shift back and forth between on-campus and online teaching and, as it turned out by September 2020, an ability to do both online and on-campus teaching simultaneously. This latter form of teaching soon became widely known as the Hy-Flex teaching model (Beatty, 2019), although its pioneering design had luckily been around for quite some time before the pandemic (Beatty, 2006).

Even though the Hy-Flex design was readily available, its application faced significant challenges in the intra-pandemic landscape. The main challenge was to optimise teaching resources within three constraints:

1. Legal: socially distancing students with no less than 2 metres between them, which was later relaxed to 1 metre. However, even a 1-meter social distancing requirement quadruples the demand for room capacity. Under the assumption that any occupied seat in an on-campus classroom is surrounded by empty seats from all sides, which roughly approximates a distance of 1 meter to the next occupied seat, there will be just 25 students in a room of 10 rows and 10 seats per row – while its capacity is 100. Therefore, planning on-campus delivery for all students meant quadrupling the demand for teaching space. Here is where we hit the second constraint: physical capacity.
2. Capacity: campus space to accommodate a socially distanced classroom. No campus is infinitely large. Although there would normally be some capacity slack at most campuses, many were busy even before COVID. Social distancing then boosted the demand for space, but that spare capacity was not readily available. Even if it was, we would have hit the third – and perhaps the most binding – constraint: staff time.
3. Time. Assume for a moment that the physical capacity constraint was not binding but the legal one was. Then, delivering entirely on-campus teaching in the Autumn of 2020 would have meant quadrupling the teaching load of colleagues. We could not realistically plan to quadruple the teaching load of any colleague, even though some working time inflation was perhaps unavoidable. To prevent rapid escalation of demand on staff time, we needed to identify the scalable elements of our existing teaching model.

The above constraints were hitting in an environment of uncertain enrolment for the upcoming 2020/21 academic year. Recognising the constraints and the surrounding uncertainty, we needed to fulfil the following tactical goals:

1. to the extent possible, protect colleagues from a blow-up in workload,
2. fit into the room capacity constraints, and
3. keep social distancing.

These goals were placed within a broader context of rapidly changing graduate job market. While we needed to repurpose our teaching model for the intra-pandemic student, we also needed to strategically formulate a knowledge and skill set for the post-pandemic graduate. The following part summarises how we met our tactical and strategic goals.

II. The essence of the teaching and assessment redesign

To achieve these goals, RHUL Econ has set up a response team in April 2020. The goal of the team was to redesign our teaching model for the Autumn of 2020. By the end of May 2020, we came up with a flexible teaching model (FTM), which addressed the topics of module design, teaching interface and student experience. Our priority at the time was to fit the module design within the above 3 constraints. To do this, we worked on two design elements: course (module) content and assessment mix. My view of those design elements can be summarised as follows:

2.1. Content

Guided by the evidence of the effectiveness of backward design (Childre et al., 2009), we have adopted it as a method to plan our weekly teaching and assessment activities. Backward design subjects teaching and assessment to the course learning outcomes (Jones et al., 2009; Wiggins & McTighe, 2005).

Using backward design in our context meant reformulating the course learning outcomes (LOs) in the first place. It also meant studying how the adoption of a hy-flex teaching model would enable the production of the newly formulated LOs. The adoption feasibility depended on two dimensions: class size (small, large), and level of teaching (introductory, intermediate, advanced). No template was ideal for all modules, but any template could use the following timeline of events for each module:[1]

1. Update the list of LOs for each module for a post-pandemic world;
2. Split the 10-12 topics in a module into approximately 30-40 conceptual units (or nuggets): shorter elements of each module centred around standard basic or more advanced concepts.
3. Produce short videos for each of the conceptual units. The videos explain the basics of the concept and offer some of its practical applications.
4. Match the LOs to class activities associated with each conceptual unit. While matching, we recommended that colleagues used Bloom’s revised taxonomy (Krathwohl, 2002).

Throughout the discussions, we agreed that the overall goal of designing activities in each module unit was not only to make students remember and understand the material but also achieve higher-level learning outcomes. Only then would students be able to transfer the skills to solve non-standard problems, come up with novel topics, and revisit old concepts and relationships for a challenging post-pandemic job market.

Naturally, content redesign was only part of the internal discussions. The other significant part revolved around assessment. Our thinking about assessment at the time is summarised below.

2.2. Assessment

Economics assessment has long been based on sitting exams. However, such assessment structure is prone to three weaknesses:

1. it is not resilient to major disruptions;
2. it may be favouring students from specific demographic groups; and
3. it could emphasise low-level cognitive skills, which is insufficient for bridging attainment gaps.

To address those weaknesses in the traditional assessment mix, we needed a new approach to assessment design. Based on the theory of constructive alignment (Biggs, 1996; Biggs, 2003), our approach managed to address all three issues simultaneously.

To build resilience in the assessment system – i.e., capacity to withstand continuous disruption and move back and forth between online and on-campus teaching – we shrank the share of final exams and midterms to a total of 40-60% of the mix and shifted the bulk of new summative assessment to continuous assessment. In addition, we spread the assessment mix along the Bloom’s taxonomy within each teaching week. This reallocated the burden away from the final exam into in-year assessment in a way that builds transferrable skills for job market excellence along with academic knowledge.

Some colleagues argue that doing a final exam may prove unnecessary at all, especially if the module LOs have been already achieved during the term (Burnett and Paredes Fuentes, 2020). To test this proposition, we experimented with removing the final exam from the introductory-level Mathematics for Economics. In 2022/23, we extended the experiment by removing the final exam in further two advanced undergraduate modules. The results from this experiment are being summarised in Stankov (2023).

On the face of it, reducing the share of the final exam meant raising the number and the share of in-year assessment. In turn, having more activities to mark would have blasted the time colleagues spent on marking and producing feedback. However, that need not be the case in practice. We were looking for a way perform a new set of marking and feedback activities without escalating the marking workload.

Preserving the marking workload meant that not all assessment would be marked by a human. Lower-level skills, such as memorisation and understanding, could be tested automatically using multiple choice questions (MCQs) in weekly quizzes. This shifted resources to producing more meaningful feedforward and feedback on high-level skills.

Shifting the burden from exams to in-year assessment was only a partial solution. We also needed to refocus assessment away from repetitively testing low-level skills into a far more diverse assessment mix. The goal of this new mix was to deliver a more balanced assessment structure along the Bloom’s (1956) pyramid and its more contemporary iterations (Krathwohl, 2002). The elements of our assessment mix testing higher-order skills were typically essays, projects, and some homeworks and problem sets. As producing this individual-level feedback can be more time-consuming than marking an offline exam, scalability was key: We needed a device to ensure students could receive timely feedback on their ideas both before and after submitting assignments.

We discussed three scalable devices, of which perhaps no more than two are implementable simultaneously within any given module:

1. Online feedback sessions with the course leaders, which effectively acted as an online office hour. This device added pedagogical value by repurposing the traditional office hour for higher-order feedback and feedforward. The feedback sessions complemented significant parts of the lectures and the seminars and offered a scalable, cheap complement to seminars and lectures. They were also timetabled for increased visibility.
2. Online synchronous seminars. With this device, both students and the module leaders participate in a feedforward-and-feedback loop, and the feedback goes student-to-student and teacher-to-student. All large modules had online seminars, and all advanced modules had a hy-flex delivery. When all teaching moved online in the Spring 2021, some module leaders experimented with breakout rooms. This has produced unexpected boost to learning and student satisfaction, as break-out rooms proved incredibly interactive at a time when human interaction was in short supply.
3. Regular departmental conferences (RDCs): a forum for creation and application of ideas in all modules we teach. The RDCs were designed as weekly, bi-weekly or monthly events, where students discuss the higher-level skills they have got from each course by that time. All students in a certain module were supposed to attend the RDCs. The key with this device was to bring senior academics on board to suggest avenues for extensions of student ideas. They could be student presentations, with on-the-spot peer-reviews by students, lecturers, and guest-professors or simply a forum for discussing the next assignment. The RDC device was not adopted in the end but could still be suitable in a variety of degree programmes where scalability is key, e.g. where courses are delivered entirely online.

Our efforts were not only directed at improving student performance in general but also targeted improvements across our key demographics. We wanted to redesign assessment in a way that bridges attainment gaps. We thought high-stake written exams were ill-suited for bridging attainment gaps, as they test the skills on one end of the skills pyramid only. We did redesign our exams to improve their skills mix, but much of the redesign effort was to broadly align our assessment mix with the course learning outcomes.

As the model was scalable across space and time, it could save significant resources to any department after the sunk costs have been internalised.

2.3. Hy-flex model summary

Ultimately, the RHUL Econ Flexible Teaching Model (FTM) took the following shape:

Note that using the traditional teaching model in an environment of social distancing would have boosted the staff contact hours from 2 hours per week per module to an average of 8 hours per week per module (even if the university could afford the space). However, now that the previous lecture of 2 contact hours has become a 1-hour lecture, quadrupling the demand for teaching space in an environment of social distancing would require only doubling the staff teaching time. This is a massive inflation of workload by all means, but is significantly lower than what the workload would have been in the absence of the remodelling exercise. In addition, scale economies can be further achieved if the small-group seminars are combined in an online session.

Based on the above, we can argue that the redesign has benefited students, as well as colleagues relative to a counterfactual of no change. However, it involved some significant implementation costs, which are reviewed next.

III. Implementation costs and expected benefits

Redesigning the RHUL Econ teaching model involved significant sunk costs. It was also expected to produce some long-term benefits. The costs involved were related to:

1. producing the content of the 1-hour weekly videos;
2. producing and uploading the automatically marked continuous assessment (ideally with embedded feedback);
3. providing the right equipment to produce the videos and the live classes;
4. training staff into using the equipment.

Content creation is perhaps the costliest of the above elements. There are technology aspects of content creation, as well as purely substantive aspects of splitting the naturally flowing 2-hour lecture content into an hour of videos and an hour of live lectures.

In addition, a mechanical split of the lecture into two hours of content would not fit into the pedagogical purposes of the model. Therefore, the videos were rather planned to deliver the fundamental blocks of knowledge for each week of teaching, while the live lecture was designed to upgrade the fundamental pieces with more applied elements, or go one step further: question the wisdom of the fundamental knowledge by evaluating its usefulness and thinking about ways to extend it (in more advanced modules).

In the background, we set up a dedicated team responsible for researching the available technologies, sourcing them, and training staff if necessary. RHUL Econ opted for delivering hybrid classes through MS Teams. In a live class, colleagues could choose to use a visualiser or a graphic tablet as a white board. Some combined the use of a graphic tablet with additional classroom technologies, e.g. Mentimeter or Socrative, to liven up the classroom. When MS Teams rolled out their break-out capabilities for the Spring term of 2021, colleagues embedded them into the classroom design with marked improvement in student engagement and satisfaction (Stankov, 2021).

Our hy-flex teaching model was superior to the traditional model for several reasons. First, its design has been guided by evidence of teaching innovation that works (Petersen et al., 2020). Second, the 3-step weekly delivery is simpler to communicate to both students and staff than alternatives based on multiple preparation, assessment and closure stages on a weekly basis. Third, the FTM enhances transferrable and employability skills. A job market hungry for quantitative, digital and adaptability skills will necessitate a shift from a traditional model to one based on a mix of low- and high-level thinking skills, where priority is given to critical thinking, analysis and evaluation. It is those higher-order skills that could carry the highest returns to education for our graduates.

The expected benefits were not only pedagogical. As the model was scalable across space and time, it could save significant resources to any department after the sunk costs have been internalised. To enjoy the scale economies, however, any department needs to preserve the scalable elements of the model for as long as possible. The longer the model has been adopted for, the larger the gains from its implementation. This is because the implementation costs are sunk in nature but are spread over larger number of students and iterations over time. Therefore, I consider the model particularly suitable for application at larger departments and ones where face-to-face teaching is not a major selling point. However, the RHUL Econ model is not scalable across all economics departments. The next section elaborates.

IV. Wider applicability and implementation risks

The hy-flex teaching model presented here is hardly unique. Many departments have found similar solutions by optimising their resources along very similar constraints. The crucial design advantage of the RHUL Econ FTM, however, is its simplicity. By splitting the traditional lecture into two new parts both of which allow for interaction, the FTM offers a way out of the physical capacity constraint, and a gradual disinflation of staff workload. The workload was still higher than in pre-pandemic times, but in a more managed fashion than it would have been without the model. Therefore, I consider the FTM a possible solution to the needs of any Economics department facing pandemic constraints or non-pandemic constraints of similar fashion to the ones described here.

However, going beyond the short-termism, any department would face the following issue: How sustainable is the flexible teaching model? Do we want to perpetuate it in full or in part after the pandemic, or completely revert to the old model? The answer to this question depends on its implementation risks, of which there are several.

First, colleagues would not automatically support the new model. Creating content for asynchronous delivery is costly and if they cannot recover the lost hours over the next few years, no rational colleague will volunteer into this model. Therefore, to minimise this risk, a department needs to think of this sunk content creation cost as an investment, and reward the colleague with a ‘break-even’ amount of time on the module. For example, in an optimistic scenario a weekly content creation exercise generates switching costs of roughly 8 hours but saves only 1 – the one cut from the live lecture and swapped for videos. Then, each member of staff needs to teach the same module – using the same teaching model – for more than 8 years to break even. In many cases, allowing for such break-even period would understandably be infeasible. Therefore, a major implementation risk is that staff buy-in would optimally be low.

Second, the model was not highly popular with students either. Despite the known benefits of flipped classrooms (Roach, 2014) and hy-flex delivery (Binnewies and Wang, 2019) for student satisfaction, our students felt that teaching was not social enough. This was contrary to our expectations and despite us creating more dedicated access points to staff than ever before. Coincidentally, the National Student Survey (NSS) scores of the Department temporarily dropped relative to the pre-pandemic years.

Third, the model was more suitable for larger classes, and ideally for classes that were scalable online. If the blend of online and on-campus programmes and degrees continues beyond the pandemic, and multiple access and exit points into and from higher education becomes the norm, then this model can provide a blueprint for design of such blended degrees. However, if a certain programme is designed to cater to the needs of a small number of students taught predominantly on campus, then there is a risk that the FTM does not pass the cost-benefit analysis.

Despite these implementation risks, RHUL Econ has kept several features of the model, which we think would pass the test of time. Those are presented next as a reflective conclusion.

V. Conclusion: What have we learned?

The flexible teaching model adopted by the RHUL Economics Department in the Summer of 2020 was a response to the COVID-19 pandemic based on combining hy-flex teaching with a flipped classroom approach. Its adoption addressed simultaneously the ongoing legal, capacity and time constraints. By nature, those constraints were only short-term. 

The model was successfully applied, it has got us through the pandemic, and has achieved its short-term tactical goals. Because of its simplicity, the model has got some attention from other departments across RHUL. It was also discussed at several internal and external education conferences. It is applicable across larger economics programmes (degree courses) and its benefits unfold over longer periods of time. However, its significant sunk costs present a serious risk of reverting back to the old model before its benefits begin to outweigh its costs. The risk of reversal would be higher at departments who fail to plan for a longer-term disinflation of staff workload as a compensation for the short-term boost in preparation time.

Three years after its adoption, we can begin to identify the features of the model that deliver on its strategic goals. The main strategic goal of embedding higher-order cognitive skills has been achieved, and we have kept most of the model features that develop higher-order thinking. The key among them is the increased emphasis – at a programme level – on application, critical thinking, evaluation, and creation of new knowledge.

We have also preserved the ability of colleagues to reuse their weekly videos and teach a one-hour lecture plus a dedicated feedback hour, instead of two full hours of lecturing per week. At the same time, new colleagues on existing modules were free to switch to two hours per week if the single hour was deemed insufficient to deliver the learning outcomes. Keeping the foundational hour of video lectures enhanced our ability to deliver higher-order skills in the live lecture or spend more time feeding forward into upcoming assessments. This feedforward time proved highly popular with students, who normally find the level of feedback and feedforward insufficient. As a result, this provided a welcome boost to our post-pandemic student satisfaction scores.

Another good practice from the pandemic that we deem sustainable is lowering the access costs to staff for students. We have preserved one hour of online feedback and office hours for increased visibility and improved access. While this makes students better-off, it does not increase the costs for colleagues, and is therefore Pareto-improving.

We have also kept a vital feature of our model: its embedded resilience. Take the transportation strikes in and around London in the Autumn and Winter of 2022/23 as an example. They were acutely felt by both our commuting students and colleagues alike, but we were quickly able to switch back to a hybrid or fully-online mode for the strike days. This minimised the disruption to teaching and learning without the need to make up any lost classroom time. Pandemics and other types of disruption prove recurrent phenomena. Then, a teaching model fit for the post-pandemic landscape needs a strong hy-flex teaching capacity to boast resilience not as an add-on, but as a basic feature of the model.

Finally, an important skill we all took from the pandemic was how to subject teaching and assessment to the learning outcomes from a degree. As a result, any new post-pandemic assessment would be routinely undergoing a check against course learning outcomes. The integration between learning outcomes, assessment design and teaching and learning experience is not yet complete. Still, it offers a welcome extension to the fundamentally positive changes we did at RHUL Econ to adopt a hy-flex teaching model for the COVID-19 pandemic and beyond.

Notes

References

Beatty, B. (2006). Designing the HyFlex world - hybrid, flexible classes for all students, Proceedings of the 2006 Association for Educational Communication and Technology International Conference, Dallas, TX.

Beatty, B. (2019). Hybrid-flexible course design. EdTech Books.

Biggs, J. (1996). Enhancing teaching through constructive alignment, Higher Education 32, 347–364. Retrieved from: https://doi.org/10.1007/BF00138871 .

Biggs, J. (2003). Aligning teaching for constructing learning. The Higher Education Academy. Retrieved from: https://www.advance-he.ac.uk/knowledge-hub/aligning-teaching-constructing-learning .

Bloom, B. S. (Ed.) (1956). Taxonomy of educational objectives: The classification of educational goals, Handbook I, cognitive domain. New York: Longmans.

Burnett, T. and Paredes Fuentes, S. (2020). Assessment in the Time of Pandemic: A Panic-free Guide. Economics Network Ideas Bank. https://doi.org/10.53593/n3298a

Childre, A., Sands, J. R., Pope, S. T. (2009). Backward design. Teaching Exceptional Children, 41(5), 6-14. https://doi.org/10.1177/004005990904100501

Jones, K. A., Vermette, P. J., & Jones, J. L. (2009). An integration of "backwards planning" unit design with the "two step" lesson planning framework. Education, 130(2), 357-360.

Krathwohl, D. R. (2002). A Revision of Bloom's Taxonomy: An Overview, Theory into Practice, 41(4), 212-218. https://doi.org/10.1207/s15430421tip4104_2

Petersen, C., Baepler, P., Beitz, A., Ching, P., Gorman, K. S., Neudauer, C. L., Rozaitis, W., Walker, J. D. and Wingert, D. (2020). The Tyranny of Content: “Content Coverage” as a Barrier to Evidence-Based Teaching Approaches and Ways to Overcome It, CBE—Life Sciences Education, 19(2). https://doi.org/10.1187/cbe.19-04-0079

Roach, T. (2014). Student perceptions toward flipped learning: new methods to increase interaction and active learning in economics. International Review of Economics Education, 17, 74–84. https://doi.org/10.1016/j.iree.2014.08.003

Stankov, P. (2021). Tying the classroom together: how to create the 'best online teaching experience' through an integrated hybrid classroom. A case study prepared for the RHUL College Excellence in Teaching awards. Retrieved from: https://intranet.royalholloway.ac.uk/staff/teaching/teaching-learning/tying-the-classroom-together.aspx

Stankov, P. (2023). The death of exams? Grade inflation and student satisfaction when coursework replaces exams, Mimeo, RHUL Economics Department

Wiggins, G., & McTighe, J. (2005). Understanding by design. ASCD Publishing