Teaching labs for statistics classes is one of those labour intensive, not very much fun, teaching jobs. In Bangor, we enjoy 300+ students a year studying psychology, and teaching the students to use SPSS to analyse data is one of those difficult and thankless tasks no one really wants.
For many years, we divided our numbers up into groups of about 50, and then repeat taught six sessions for 90 minutes for about eight weeks each semester over three semesters. Even with an average of three postgrads helping out in each lab session, the results were never very positive – the most able were bored, the less able were lost, and the middle felt pushed, but kind of got it.
My colleague, Mike Beverley, was responsible for teaching the labs across both the first and second years while I taught the first year classes. The topics covered ranged from simple data entry to complex ANOVAs and factor analyses. Students were set assignments, analysed data, and turned in regular work. The experience was not very enjoyable for either the students or the teachers. Our satisfaction ratings from the students was low when it came to the labs (“Burn every copy of SPSS on the planet!” was typical). We needed a new model.
We initially decided to podcast Mike’s first session each week, and then he could ensure that the presentation to the students was at least consistent, and he didn’t have to repeat himself endlessly. Podcasting was new then (spring of 2005), so we were just trying things out. At first we recorded audio podcasts, but after a few weeks, we recorded the screen capture for the students. By chance, we stumbled onto a model that really worked. In the labs, we would usually provide about five minutes of instruction, and then let the students work at it for a few minutes before introducing something more. As a result, all of our podcasts were about five minutes long, demonstrating how to carry out a procedure with a voiceover. The initial fumbling about was successful enough that we decided to prepare podcasts over the summer to cover every topic we taught for use the following year.
In the Autumn of 2005, we scheduled the labs, employed the postgrads, and demonstrated the podcasts to the new, incoming students. To our surprise, no one ever came to another scheduled lab. I fib here – there were about six students who insisted on coming every week. After about three weeks, we rolled the six into a single session, and let the rest of the students know that we were only going to be in the lab for that single 90 minute session. If they had question, they needed to come along then.
The students learned SPSS – better than they had in previous years. Their feedback in the module evaluations was uniformly positive about learning SPSS that way, and we changed the way we did things (we still use the same basic model six years later).
The savings from this have been great for the Department. Teaching the traditional labs went something like this:
- 9 hours/week per instructor
- 4 instructors = 36 hours/week
- 8 weeks teaching across 3 semesters
- 864 hours/year of instructional time
- Stats support surgeries for 3 hours/week across 15 weeks involving 2 or 3 people
- Approximately 75 hours/year
- Total of about 940 hours year teaching and supporting stats
Using podcasts for instruction, the cost went someting like this:
- 22 weeks with 1 hours of support available per day (1 person currently)
- 110 hours
- Savings of 830 hours
- Estimate about 3 – 4 hours per lesson
- 2 or 3 podcasts per lesson
- 23 lessons
- 70 podcasts in total
- About 100 hours in total to make the initial podcasts
Massive savings overall (830 hours less development time), and the podcasts are reusable. The instructor was happier (for a time), the students were happier. The students have the podcasts available throughout their entire undergraduate programme so they can refer to them anytime. They have control over their learning,and use the labs when they choose to do so. A real win – win solution!
The single biggest problem has been the updates of the programme (SPSS). This has meant that we have re-recorded the podcasts twice in six years – not too bad an investment given the long term benefits.
I think there were a few things we did (and continue to do) right when we made the teaching podcasts. They were:
- Have an expert in Stats & SPSS teaching make the podcasts (at least the first time)
- Keep the podcasts short (5 – 10 minutes)
- Don’t obsess about quality
- Be prepared to release updates quickly if there is a need for clarification
In the area of teaching and learning, given the promises of efficiency and performance that technology has alluded to over the years, it is nice to see something that turns out to really work – along with some quantifiable evidence to illustrate just how well.
We started this way back in 2005 – I just haven’t put this out there since. I have presented it at a couple of conferences, but not put it out there where anyone could refer to it, so here it is.