Tuesday, July 3, 2018

Engagement surprises in data and statistics

In this post, I'm talking about some developmental work I'm doing on engaging data/statistics teaching. But I think the interesting part is the observations made by my 9 year old son and the discussion with my friend that was sparked by the whole activity.

Some colleagues and I at QUT are currently working on a couple of related projects simultaneously. One of these is part of our new Bachelor of Education development, which (assuming various approvals go through) will be live next year. There we are looking to refresh the way we and our pre-service teachers look at curriculum, pedagogy and assessment for mathematics as well as the study of the mathematical ideas themselves. The other project is related to continuing professional learning in mathematics and teaching mathematics, directed at teachers already in the field.

As part of both of these pieces of work, I'm looking to make our mathematics classrooms more active and engaging, and to incorporate more inquiry and outdoors activity into the learning. At the moment I've got my head in the junior secondary probability and statistics world, and I'm toying around with some simple activities that will get my students coming up with questions, and then collecting, analysing and visualising their own data.

A few days ago I asked my son if he would help me to try out some ideas and surprisingly he was keen! I suggested that we could do some sporty things such as throwing balls at targets and timing runs of various distances. He improved on my suggestion by turning the ball throwing into a real question for inquiry. More or less, he suggested we look at whether or not there are differences in accuracy when you throw with your left and right hand, and with or without pointing the other arm at the target. The target was another ball some 5 or so metres away, so our activity boiled down to something similar to the situation involved in lawn bowls. That is, he threw a ball, then another, then another, each time trying to get it to stop close to the target ball. After three throws (we had three balls), we would measure and record the distance from the centre of his thrown ball to the centre of the target ball. The screen cap of a tweet I made, shown below, includes a couple of photos of my son throwing (here with his non-dominant hand) and then measuring one of his throws.

We recorded distances in Excel for 30 throws for each of: right arm-no pointing, right arm-pointing, left arm-no pointing, and then unfortunately ran out of time as it got dark and cold! After the first set of 30 throws, I showed him how Excel could automatically calculate a mean (he's familiar with "average" as being the sum of scores divided by number of scores) using "=average(data)" as well as the maximum, minimum and the range. We talked about how the minimum distance might reflect his best throw at the target but that the range gives an idea of the spread of all of his throws. We then got talking about whether he got lots of throws close to the target, or not many, or...? I tossed up showing him a histogram but eventually went with a boxplot (again using Excel's built in capability) as I wanted to compare these results with his other sets of throws. With the boxplot, we talked about the quartiles and what they represented (eg the best 25% of his throws and where they landed).

Anyway, moving on, we started to do the remaining trials. But before we did, for some reason I copied the autocalculated average across to where it would be needed for right arm with pointing, and the left arm throws. As he started to do his right arm pointing throws, and record his data on my laptop, of course the average down the bottom of the column was changing with each additional data entry. Then a cool thing happened. He did one particularly bad throw, a distance well above what was currently his average, and made a really great observation which I quoted in the tweet below.

That in itself was great - my 9 year old makes an interesting and fairly insightful observation as a result of our activity. Awesome. But then the tweet that followed led to some more interesting conversations online with my friend Rob.

Rob and I are both cricket lovers, so this conversation took a turn towards cricket and cricket averages, and the story of the great mid-20th century cricketer Sir Donald Bradman. For the non-cricket people, "the Don" is probably the greatest cricketer of all time and his stats have even been analysed in various ways that put him as a sportsperson well beyond the likes of Jordan and Gretsky when compared with their respective sporting peers. Anyway, one of the great Bradman stories is that when he batted in his last innings, he needed to score only 4 runs to finish his career with a batting average of 100. Unfortunately, he was dismissed for 0 and his average for ever after has been 99.94.  This was the topic of Rob and my discussion (see tweet grab). Also, check out the box for the slightly odd calculation for cricket batting averages.



Aside: The oddity of cricket batting averages

In cricket, a player's batting average is calculated according to the formula
where a dismissal means the batter was out. So for example, if I batted 3 times, and scored 50 before getting out, then 20 before getting out and then 30 but did not get out, my average would be 50=(50+20+30)/2. Note that we divide by two because I only got dismissed twice even though I batted three times.

I had never thought of this, but had kind of always assumed that Bradman was increasing his average to 100 in that last innings. Rob's question made me think about this and of course, realise that this wasn't the case. Before that final innings Bradman already had an average over 100. The 0 or "duck" in his final stint at the crease dropped him down (similar to my son's observation), and enough so that his average fell below 100.

This got me mucking around of course - I went and downloaded the data for each of Bradman's 70 innings and played a little. One of the plots I put together was that below, showing his average varying over time. At one point he had an average of over 112 runs per dismissal. Phenomenal. And interesting statistics fodder!


Wednesday, February 14, 2018

Traditional mathematics assessment... but enhanced online

Online mathematics assessment... how do you do it?

Most of the time in mathematics when we hear "online" and "assessment" we're thinking or talking about multichoice quizzes in a learning management system, or perhaps the more mathematically sophisticated options offered by WeBWorK or Numbas. That's all great, but it's not what I'm talking about here. This post is about:
  • students submitting handwritten or typeset work online
  • academics grading and providing feedback on that work, again, online, and
  • returning graded student work to them via online means.
Until 2016, my University mandated the use of a centrally managed assignment submission system. The system was a physical location where students could print out (if needed), hand in, and obtain proof of submission for assessment items. While I'm sure some academics still asked students to submit in class, by email or via "drop boxes" located in departments, none of this was really "allowed". In 2016, the submission system was closed in preparation for a compulsory online submission process.

Now in theory, this is great. Online submission is trackable and proof of submission is fairly easy. Plagiarism detection is significantly automated through, for example, turnitin. Typed feedback and rubric-based grading is also often built into these systems. But what about problem/solution style quantitative disciplines like mathematics, physics and engineering? My colleagues and I found great difficulty in adhering to this new system given that we wanted students to submit worked solutions to mathematical problems (often entirely symbolic/graphical representation), to handwrite feedback to them (again in symbolic/graphical representations), and to get this feedback to them as quickly and seamlessly as possible.

There are certainly issues here, but not insurmountable ones:

  • Handwritten work can fairly easily be scanned to PDF through photocopiers or "scanned" via smartphones. Online submission is then feasible. ADDED PROBLEMS: File formats, unfamiliarity with technology.
  • Handwritten feedback can be provided by printing out PDFs or by using PDF annotators on smart devices/tablets. ADDED PROBLEMS: Printing costs $$$ and is environmentally troublesome. Tablets/devices are costly and generally speaking not owned by all academics. The ability to use the PDF annotators requires training.
  • Feedback can be returned to students by re-scanning the printed copies or transferring annotated files back to the students. ADDED PROBLEMS: Scanning takes time ($$$). Transfer of files is not seamless and again takes time ($$$).

One of the most successful ways to get over most of these issues, that I've found, has been to use Microsoft's OneNote with its ClassNotebook add in. At its heart, OneNote is essentially freeform note taking taking software. It allows both text based and inking input (ie using a pen on a tablet), as well as the capability to import the contents of a file (say a PDF) into a note space. In the maths classroom, this is great because these three things allow students to input their mathematics in authentic ways: handwritten straight onto the screen, typed (just like in MS Word docs), or handwritten and scanned to a PDF and imported to the notebook. So it seems that one side of the story (the student) is catered for.

But what about the academic and the feedback process? Well that's where ClassNotebook comes in. The addin is in some sense a way to provide learning management system-like capabilities from within the OneNote notebooks you might want to create, as well as the sharing of notebook spaces. There's lots of things that can be done here including simply co-authoring notes, communicating synchronously and asynchronously, locking spaces, and so on. But back to the issue of online assessment in maths, here's what I did.

First, I was able to share a OneNote notebook with all of my students so that they could see the same notebook as me. This can either be done manually, or if your LMS is setup appropriately, by linking to that LMS and the contained classlist. I had notes and so forth in the notebook that I shared with the students. They could see these note pages, make copies of them and then edit them as much as they liked (I retained a clean copy). Students each had their own personal spaces in the notebook that only they and I could see. We could interact through these as well. I also had my own hidden space where I could keep notes etc.

Now for the key part: I could share an assessment item handout with my students and automatically set up a space where each student would attempt the assessment. When the due date and time came around, I could grab each of these (electronically speaking) and keep what amounts to a timestamped copy of their work, which I would then edit, mark, annotate etc. I could do this by typing, or because I have a tablet by inking straight onto the work itself. If I wanted to I could even make audio and video annotations if that made it easier to explain something. When I was ready, I could return a copy of this to each individual student's workspace with a couple of mouse clicks.

Fantastic! But this is just one idea. I'm really interested to know what other people are doing in the electronic/online assessment space. Let me know!

Class response systems and (re-)shaping the lecture

This post is kind of in response to Amie's tweet above, and kind of something I probably should have done ages ago. Also, I make no promises about this being novel or interesting!

In 2015 I started running my first year introductory engineering mathematics course in a fairly structured, blended (some would say flipped) manner (see references below for further detail). One of the big parts of that was trying to extract the "process-heavy" elements from my f2f lectures, moving them to videos and out of class activities. This extraction allowed for more interactive learning/more extended/broader learning in the f2f classes. But it also meant I needed ways to ensure that people were progressing with the outside of class activities AND getting something out of them.

What I'll describe below is essentially one of the ways I went about this checking or diagnosing if you will. In the end, it became a way for me to both check in on students' progress and understanding as well as to really shape the f2f learning and teaching to suit the students present at the time. So here we go.

---------------------------------------

Prior to each f2f lecture class, I would create 3 or 4 multi choice questions on GoSoapbox which related to the process/content videos and activities that I had asked students to work on over the prior 7 days. Sometimes these would be process questions themselves (eg carry out this operation, differentiate this function etc), while sometimes they would address understanding of concepts. Generally though, they were not overly deep questions and meant to be something that be completed in around a minute or two each for a prepared student.

Prior to commencing the lecture, the GoSoapbox details were provided on the screen and students were directed to attempt the questions using whatever device they had with them. I would allow around 10 mins (plus what ever pre-lecture time was available) for students to attempt the questions, before calling the class to a stop. At that point, I would project the results for each question - GoSoapbox provides percentage bars for poll-type questions for example. Here's two examples.



Now it's at this point that the results would determine what was to come! I usually planned for an "ideal world" scenario. In that world, the majority of those present would successfully complete the problem. I would show the results for a question, indicate the correct response, explain away any distractors, and provide references to materials etc for those who had troubles. Then we would move on to extension activities or interactive activities etc, building on the pre-lecture work.

Of course, the world wasn't always ideal! It's in these cases that things are able to get interesting and learning opportunities raise their heads. For example, I could simply go about presenting a worked solution to a question live in front of the class. That's ok, sure, but there is more that can be done. For example, when a significant percentage of the class answered a question incorrectly, we would focus on their misconceptions and attempt to unravel them and reform a better/more correct conception. Or if there were two responses that students favoured (one correct, one incorrect), we would look at the differences and why one was correct and the other incorrect.  Sometimes we even got interactive and students would discuss with their neighbours how they arrived at a particular answer, and try to figure out who was right and why.

Regardless of what happened though, the great thing about this was that I wasn't just guessing where my class was at - I was collecting the data in real-time to know (well, close to it) where they were at. We could move on if it made sense, or stick around and iron out the creases if need be. The experience was meaningful, and hopefully useful, to those present.



Shameless plug for our papers...

Czaplinski, Iwona & Mallet, Dann G. (2016) Preparing future graduates to become lifelong, expert learners: Lessons and considerations from a blended learning engineering mathematics unit. In The Australian Conference on Science and Mathematics Education, 28-30 September 2016, The University of Queensland, Brisbane, Qld.
Czaplinski, Iwona, Mallet, Dann G., Burrage, Pamela, & Psaltis, Steven (2015) Preparing engineering graduates for the knowledge economy through blended delivery of mathematics. In HERDSA 2015 Conference: Learning for Life and Work in a Complex World, 6-9 July 2015, Melbourne, Vic.
Czaplinski, Iwona & Mallet, Dann G. (2015) Investigating the effectiveness of an ecological approach to learning design in a first year mathematics for engineering unit. In ascilite 2015. Globally Connected, Digitally Enabled, 29 November - 2 December 2015, Curtin University, Perth, W.A.