Understanding questions algorithmically: Update, I was wrong.

This is an update about my last blog, trying to apply algorithms and heuristics to exam questions. It is also a bit of a love letter to edu-twitter.

My initial thoughts were that by making my thinking explicit I would be able to teach the students an approach they could apply to all questions.

I was wrong

I was fortunate to get into some great twitter conversations about the idea with some members of the CogSciSci group. I had ignored the knowledge required to understand the deeper structure of the question. I had started from the massive assumption because the students had A and B grades in Chemistry GCSE and that I had ‘expertly’ taught them bonding last year they had grasped the concepts required and it was just an exam literacy issue. Afterall, they passed the bonding end of topic test and the end of year 12 mock.

We also debated the importance of cognitive conflict. I had always loved the idea that people hold onto their misconceptions tightly (which is definitely true) but that, as an incredible teacher, I could put them in a position where I could show them that they  were wrong, I was right and that I was an amazing teacher. You can see why this approach might be desirable

Under their advice I read Shtulman’s 2012 work on tuition vs intuition and ‘knowledge in pieces’ which is also incorporated in Gethyn Jones’ excellent blog as part of the curriculum in science symposium on twitter.  It became clear reading this that preconceptions are always prevalent to some degree and that it is more about establishing and strengthening new or improved schema than obliterating old ideas through enlightenment.

That moment when you claim to know genetics but then you sound like a Channel 5 documentary

These deep-rooted ideas have a habit of showing themselves when we least expect them. I recall a particularly embarrassing tutorial at uni where I was discussing the cloning of Hitler with another student (it was the early 2000’s). I made the massive mistake of thinking he would be a full grown adult clone for some reason, despite knowing a large amount about cloning and genetics at the time. This naive misconception had obviously stuck around in my brain waiting to embarrass me. The storage strength was so high that although the recall strength had diminished over time the idea still persisted.

So I re-formulated my approach with my class on Wednesday. I did quickly discuss and model the thinking of how to approach the question, what to ask yourself and drew some simple diagrams to show my thinking. But, what I did afterwards was much more powerful. I had made a simple worksheet which took each aspect of the content in turn and gave the students practice.

I started with identifying the types of bonding from a range of metals, covalent molecules, covalent macromolecules and ionic compounds. Then some simple questions to summarise the change in atomic and ionic radii across a period and down groups.

Once I was happy with the students answers to these I let them practice a series of questions designed to consider all the different factors that affect bonding strength. They compared metallic to ionic, various sized and charged ions, covalent to ioinic, molecular to macromolecular etc..

What was fascinating was where they got stuck. I was shocked how many struggled to get the initial task right. Even students who got decent grades in their mocks apparently had their bonding knowledge built on loose foundations. The task allowed me to spend time discussing their gaps and help close them. It also helped my most capable student as their language in their answers was sloppy and rushed, so they could re-draft, improving their use of language.

Algorthims are not the answer to the competent student

I was wrong to think people need to approach the question systematically. This is too slow for successful students in the exam. They might need to start this way when they are novices but we need to help them secure strong schema which through experience they can apply correctly by understanding the deeper structure of the question. Afterall this is why we develop heurisitic approaches, algorithms are too slow.

An apology to Mr Gove

I don’t like Michael Gove. I hate nearly everything he stands for. But I do think the science curriculum he oversaw is better than what was before. I think the more I learn about cognition and learning the more I value the modern linear specification as a preparation for A-Level (although there does need to be a viable and valued new qualification to replace the old single award for less able students). The class mentioned above all come through the legacy specification and as such could compartmentalise information and cram for individual tests. This was made painfully apparent with a particular student who was really struggling. After a few minutes trying to explain how to identify the type of bonding  I realised they were stuck at an even earlier issue, they could not recall where metals and non-metals were on the periodic table. This is probably the most convincing piece of anecdotal evidence I have to support linear and synoptic assessment in all key stages.

A thank you to Edu-twitter

About a year ago I was looking at my mortgage statement and thinking ‘can I afford to step out of middle leadership?’ and even ‘can I step away from teaching?’. It was not a good time for me. A year later and I think I’ve never been more certain about what good teaching entails, how best to run a department and what we need to develop. This is all down to the amazing community of teachers on twitter, who have shared ideas, resources and advice. It has allowed me to find like-minded individuals, consolidate my thoughts through this blog and even opened up opportunities I would never have been afforded otherwise.

THANK YOU