It was a Thursday afternoon, I was teaching Year 9 about electricity, it was raining, and there wasn’t enough equipment to go around the department.

Welcome to the reality of science teaching.

But rather than complain, I saw this as an opportunity. I had been doing some research around the use of models and their application in science teaching. This was the perfect opportunity to apply it.

Modelling in science has a rich history, with the technique being rolled out to explain concepts that may be:

* Too small for students to see (eg, particle model).

* Too abstract to see or understand (eg, electricity).

* Too slow to demonstrate (eg, Lamarck vs Darwin in evolution).

* Too difficult to immediately understand conceptually.

Now, most science teachers have many models in their repertoire. We have our favourites, our fail-safes and our last resorts. But how many of these can we honestly claim to know are effective? How many are, in fact, not aiding explanations at all?

I admit to occasionally slipping into some bad habits with modelling, particularly over- relying on tenuous models to explain difficult concepts.

For example, one of my classes had previously studied genetics and, when tackling Punnett square questions, I used modelling (badly) to explain dominant and recessive.

What I did was this: I created a real-life Punnett square and used students to represent each allele; we then divided the room up into the four quadrants and asked students to be physically moved into the four genotypes. I selected students to represent ‘‘dominant’ and “recessive” alleles based on either their physical stature or personality traits. I carefully explained the key terminology such as “homozygous”, “heterozygous” and “genotype”, and felt from assessment for learning that students had understood the model.

Fast-forward a week or so to the end of unit assessment, and a question featuring a Punnett square: “Give the reasons why all of the children had brown eyes.”

Here are some of the answers: “Because it is stronger”; “Brown bullies blue”; “It wins.”

Clearly, the model had failed. And clearly, I needed to do some research on modelling.

It turned out that in order for my students to access models, they needed to understand the nature of models, their relation to the concept being studied and why scientists use them. Grosslight et al (1991) describes three levels for understanding the “nature of models”.

Beginner: “I think that models are a direct copy of reality and don’t see how they differ from reality.”

Intermediate: “I understand that models are not direct copies of reality and I know that some parts of the model are different from reality. I understand that models are used to help me develop my scientific understanding.”

Expert: “I know that several different models can be used to explain different aspects of an idea or object. I understand that models have strengths and weaknesses, and that existing models can be changed and improved. I know that models can be used to test ideas and are created for specific purposes.”

My class had clearly been stuck on “beginner” level. How do you move them up?

One method to develop students’ understanding of models is to give them experience with a wide range of model types, then ask them to evaluate their effectiveness. My reading of Treagust (1993) also suggested that building on the models would be beneficial.

So, what did I do that rainy Thursday to teach Year 9 about electricity? I rolled out the water circuit model. This model uses a pump to represent the battery, a turbine to represent the light bulb and water pipes to represent connecting wires.

Instead of just demonstrating the model and getting on with the lesson, students were asked to discuss the features of the model and compare the similarities and differences between the model and the electrical circuit.

Students were able to explain the similarities between the model pump pushing water around the pipes and the battery pushing electrons around a circuit. They were able to articulate how pipes carried the water in a similar way to wires carrying electricity and the importance of a complete circuit.

Crucially, though, when considering differences, they were unable to relate the work of a turbine to that of a bulb, ostensibly due to their lack of exposure to turbines.

This highlighted another area of Treagust’s work: if the model is more unfamiliar than the concept it is trying to explain, it may hinder and not help students’ understanding.

Another benefit of the model, though, came to light when a student relayed a story of the water pipes in her house cracking and water leaking. This allowed us to discuss whether this could happen with electricity and it really helped pupils to move from beginner to intermediate on Grosslight’s scale.

In subsequent lessons, students reflected on the use of the water circuit model. They considered whether it was clear or confusing, and created iterations of the model, suggesting improvements where appropriate.

The short-term impact of this was very clear in the end-of-year assessment. Students were now able to see the links between the “invisible’, the models and, crucially, the practical work leading to a quicker and deeper understanding of a key concept.

Before using a model, I now always prepare with the following in mind:

* Teaching students why we use models. Discuss the strengths and weaknesses as outlined above.

* It is important to remember that all models are limited. This is particularly prescient for new teachers, who will have limited experience in using models to explain concepts.

* If the model is unfamiliar, abandon it! The use of Christmas tree lights as a model of series circuits is now obsolete and confusing;

* Encourage your students to develop their own models and share them. Research shows that students who are involved in constructing their own models are better at critiquing models as they understand the modelling process.

Phil Naylor is assistant director of the Blackpool Research School and expert adviser, Blackpool CPD Hub, for the Teacher Development Trust