Transitioning an in-person statistics class to online
An instructor reflects on transitioning a statistics class to an asynchronous online class.
“That’s okay. A statistics class should transition pretty well to an online environment.”
I heard this statement in various forms from my faculty colleagues in the summer of 2020 as we were preparing our fall semester. This was when many of us were preparing to teach our in-person classes in an online environment for the first time.
My faculty colleagues would then describe to me in detail all of the hardships they were facing in transitioning their courses online. Meanwhile, I thought of all the videos I needed to pre-record, the coding tutorials to put together, and the online quizzes and assessments to create for my course.
This is a post that reflects on the fall 2020 semester and what worked and didn’t work in transitioning a graduate statistics class taught in person to an online asynchronous class.
Asynchronous online classes are effective for introverts
I instruct the course NR 5021: Statistics for Agricultural and Natural Resources Professionals at the University of Minnesota. The class has between 25 and 40 graduate students enrolled in it. I have a teaching assistant that grades lab assignments and holds a few office hours every week. Students are typically in their first year and enrolled in a Master’s or PhD program.
Prior to this semester the class met twice a week for 75 minutes each in a 15-week semester. I would lecture on a statistics topic, provide class activities and problem sets, and encourage students to work in small groups on coding problems using R software.
This semester I chose to deliver the course as an online asynchronous course using Canvas, our university’s learning management system. There were 21 statistical topics which were given as “lectures” in the in-person course. One or sometimes two topics were presented in each week.
Short lecture videos (typically less than 10 minutes each) focused on a specific statistical topic. Problem sets and calculations were also recorded to highlight statistical concepts. Depending on the topic, between three and eight videos were shared with students to watch at their own pace.
I instructed students to tackle this course as if they were paying off a debt: consistent work leads to a solid grade. Because of this, the format of the course required students to be diligent in completing the course requirements. Each topic had an assessment, typically a quiz or short written assignment that assessed their grasp of the material. By keeping pace with the material and consistently putting in the effort, students accumulated points that were allocated to their final grade.
As Susan Cain describes in her book Quiet, introverts are most comfortable when they have the ability to learn independently. Introverts make up at least a third of the population, but probably more in my statistics class full with science-minded graduate students.
I have no doubt that introverts have thrived in coursework that is asynchronous in its design. The ability to work independently at one’s own pace appeals to many.
Group activities don’t translate well to asynchronous online courses
While online asynchronous classes in statistics may be fitting for introverted students and the faculty that instruct them, these course designs are not likely to be as effective for extroverted students. Group activities and collaborative work is challenging to arrange in an asynchronous format.
I found it difficult to organize activities for students to work on in a group in an asynchronous course. I was not offering the opportunities for students that learn best by working in teams and are energized from collaborating and socializing with others.
But then I attended a webinar offered by e-learning specialists at my university and was basically told “If you try group work in your asynchronous course, let us know how it works for you.”
The interest in doing group work relies on the student in an asynchronous course. Faculty are constrained by getting students together at a specific time: what time works for some does not work for others in an already packed semester when students are faced with Zoom fatigue. Faculty are also not permitted to provide email addresses or other contact info for students due to FERPA regulations.
I offered one major assignment that offered students a choice to complete it individually or in a group. Six of the 35 students in my class chose to complete the assignment in a group. Despite the few opportunities for group assignments, many students organized Zoom study group sessions and group texts to talk about the course materials on their own.
Gathering data is difficult
In a statistics class, one of the best ways to engage students in the material is to use data that they collected. In an in-person class, I would often provide a Google Form that students would complete synchronously. Then, I would have some pre-written code that I would project in the room that visualizes their responses and runs some statistics.
This approach works great for showing concepts related to residuals in linear regression (after guessing how many jellybeans are in a jar), visualizing linear regression lines (after predicting how tall celebrities are), and interpreting confidence intervals (after asking how confident students were in their estimate of how many moose are in Minnesota).
The challenge with collecting and using data in an online asynchronous course is that time elapses between data collection and when the student sees the results. If I want students to provide data that I’ll use in an activity, I really need to ask for the data in the prior week’s module. The duration between data collection and results gives the concepts less impact. The student misses the “A-ha!” moment when data are collected and analyzed in the same class period.
I highly recommend anyone new to teaching statistics read Andew Gelman and Deborah Nolan’s Teaching Statistics: A Bag of Tricks. The book is full of excellent activities to add to your quantitative courses, and while most examples are fitting to an in-person course, I found many to work well in the online environment.
As one example, Gelman and Nolan describe an activity where you pass an inflatable globe of the world around the room. After a student catches it, he or she indicates whether their right index finger lands on water or land. The result is a data set that will calculate how much of the earth is covered in water. I’ve used this activity with great success in the classroom in the past. I almost scratched it from my online class before finding random.org, an online tool that drops you at a random spot in the world. While missing the fun activity in the classroom, this online tool provided us a data set to sample the amount of water on earth. A student would use the data to calculate a confidence interval of the amount of water on earth and determine if it contained the true value. (For your reference, earth is approximately 71% water.)
Office hours are better on Zoom
Before COVID, my office hours were structured like every other faculty member: stop by at this scheduled time and I’ll answer any questions you have. I ran virtual office hours for my statistics course prior to COVID, and while effective, they had relatively low attendance.
Fast forward to during the COVID pandemic and all office hours are conducted via virtual platforms such as Zoom. Office hours were the only synchronous component of my statistics class this semester, and with the help of an excellent teaching assistant, we held one hour on each weekday at different times throughout the semester. This consistent time on the schedule was effective for students that rely on routines for completing coursework.
Students use RStudio to complete lab assignments, and this is what most of the questions during office hours were about. The ability for students to screen share on Zoom made it easy for myself as the instructor to diagnose the problem with a student’s code. During in-person office hours, a student would bring their laptop and I would often squint to see their code. The student would often not keep their laptop’s brightness as bright as I prefer. These are not good combinations for a near-sighted instructor that constantly switches between taking his glasses on and off.
With Zoom, a student’s code is clearly visible on the screen. I also like that other students that are sitting in the office hours can “eavesdrop” on a conversation and listen to conversations with other students. Typically office hours in my course had three to four students at a time.
While many instructors use the virtual waiting room feature in Zoom for office hours, I don’t care much for it. Many students have the same questions and challenges with their assignment. Having multiple students in office hours at a time is a small component of peer-to-peer learning that an instructor can facilitate in a virtual environment.
Audio will rule the future of learning
I mixed in several podcasts into my course material and received positive feedback from students. This came from several emails and mid-course evaluation comments from students. I was somewhat surprised that several students mentioned concepts they heard on a podcast in written assignments that were due several weeks later.
With the increase in popularity of audio books and podcasts, incorporating course material in audio form will be needed to meet diverse learning styles in the future. In statistics and data science, there are many great podcasts to choose from. I’ve assigned listening to episodes from Linear Digressions, Not So Standard Deviations, and Data Framed, to name a few.
I would not likely have assigned podcasts for an in-person course. (Don’t students hear enough after sitting through two 75-minute lectures each week?) Learning by audio is likely to increase in popularity and there is a tremendous opportunity for growth in this area for statistics and data science classes.
I’ll probably continue the class online in the future
When in-person classes come back after the COVID pandemic, I know that many university faculty will jump at the chance to be back in the classroom. But given the success I’ve seen this past semester, I’ll likely continue my statistics course in an online asynchronous format after the pandemic.
There are still challenges that university-level statistics and data science courses will have in the future, namely how to encourage group work and collaborations with assignments and how to facilitate peer-to-peer learning. From an instructor’s perspective, there are many benefits of an asynchronous online model for teaching statistics.
Statistics courses translate well to an online environment. Well into the COVID pandemic, university instructors can no longer ask for so much grace and flexibility from students in delivering online courses. Students expect faculty to design and deliver online courses well, have consistent due dates, and provide continual feedback.
