# Uncertainty about Uncertainty in IB Science

I have a student that is taking both IB Physics with me and IB Chemistry with another science teacher. The first units in both courses have touched on managing uncertainty in data and calculations, so she has had the pleasure (horror) of seeing how we both handle it. For the most part, our references and procedures have been the same.

Today we worked on propagating error through the calculation $$\Delta x = \frac{1}{2}at^2$$ with uncertainties given for acceleration and time. The procedure I’ve been following (which follows from my experiences in college and my IB textbooks) is to determine relative error like this:

$$\frac{\delta x}{\Delta x} = \frac{\Delta a}{a} + 2 \cdot \frac{\Delta t}{t}$$

In chemistry, they are apparently multiplying uncertainty by 0.5 since it is a constant multiplying quantities with uncertainty. On a quick search, I found this site from the Columbia University physics department that seems to agree with this approach.

My student is struggling to know exactly what she should do in each case. I told her that everything I’ve seen from the IB resources I have in physics supports my approach. The direct application of the formula suggests that an exact number (like 1/2) has zero uncertainty, so it shouldn’t be involved in the calculation of relative error. That said, the different books I’ve used to plan my lessons agree with each other to around 95%. There is uncertainty about uncertainty within the textbooks discussing how to manage uncertainty. Theory of knowledge teachers would love the fact that teachers of a generally objective field (such as science) have to occasionally acknowledge to our students that textbooks don’t tell the entire story.

The reality is that there are a number of ways to handle uncertainty out in the world. Professionals do not always agree on the best approach – this conversation on the Physics Stack Exchange has a number of options and the mathematical basis behind them. For students that are used to having one correct answer, this is a major change in philosophy.

Thus far in my teaching career, I haven’t delved this deeply into uncertainty. The AP Physics curriculum doesn’t require a deep treatment of the concepts and roughly ignores significant figures as well. I talked about some of the issues with uncertainty with students, but I never felt it was necessary to get our hands really dirty with it because it wasn’t being assessed. We also learned error analysis in my experimental design courses in college, and it was part of the discussion there, but it was never the class discussion. It’s really interesting to think about these issues with students, but it’s also really difficult.

It seems that the questions that have resulted both from class and for my own understanding are exactly the style of conflict that the IB organization hopes will result from its programs. The way this student throws her hands up in the air and asks ‘so what do I do’ and managing the frustration that results is the same difficulty that we as adults face in resolving daily problems that are real, and complex.

The philosophy that I shared with the students was to be aware of these issues, but not to fear them. It should be part of the conversation, but not its entirety, especially at the level of students that are new to physics. I’m confident that some of the discomfort will melt away as we do more experimentation and explore physics models that tend to describe the world with some level of accuracy. The frustration will yield to the fact that managing uncertainty is an important element of describing how our universe works.

## 2 thoughts on “Uncertainty about Uncertainty in IB Science”

1. Andy "SuperFly" Rundquist says:

I love the line about it never being THE conversation. We take up a week’s worth of time in one of our early labs in our curriculum on it and then just hit it lightly again in the rest of the labs. In the Global Physics Department we had an interesting conversation about how to teach this, with the montecarlo approach getting the biggest support (that’s certainly what I push). What I’d like to do more of is experiments with much more repetitive data to really show true histograms. Do you do any of that?

1. I haven’t done anything with Monte Carlo, no, but it seems like a good way to convince others (and me) of what the right approach might be.The fact I didn’t jump to a computational approach must mean I’m rusty – thanks for the suggestion!