With 20 hours allocated for the IA and a lot to get done, I only have time in my course to plan one lesson for inferential statistics. In this time I want students to get a basic understanding of:

  1. how inferential stats differ to descriptive ones
  2. how to choose which inferential statistical test to use
  3. and most importantly, why inferential statistical tests are applied to data

One really simple way of achieving 1 and 3 is by doing the following quick activity.

In due course I’ll actually create these lesson materials and post them so you can just pick it up and adopt, apply, adapt it for your own lessons, but for now I’ll just give a basic outline of the lesson plan. You can see the resources on the slideshow below as well.

I give them two sets of data with the calculated mean. The example I use in the slideshow is of an experiment (that they wouldn’t do for their IA) which is how many press-ups can be done with music or no music (I like to use non-IA examples so they have to think for themselves how to apply the learning to their own studies). From the descriptive statistics you can see that the mean is 18.6 (music) vs 15 (no music). I get students to draw a conclusion based on this data.

They’ll naturally conclude that listening to music helps with workout performance. But then I show them the next slide with three numbers on the control (no music) group highlighted: you can see (on the slideshow) that they actually had more push-ups than the treatment (music) group. Now I can challenge students to analyse the data and to see if they would trust these conclusions when 3/7 of the participants’ scores are not consistent with that conclusion. We might discuss the implications of conclusions in experiments like drug trials: if you were experimenting a new drug that treated, let’s say Parkinson’s Disease, would you allow it to be manufactured and sold with these results?

And now it should be rather straightforward to explain to students the purpose of inferential statistics: they enable us to statistically calculate whether or not we can “trust” our descriptive statistics. Researchers don’t have to analyse the data just by looking, as we’ve just done, because in most quantitative studies there are far more than 7 people and sometimes in the 1,000s, so this would be impossible. We apply the inferential tests to see if our results were actually a product of the manipulation of the IV or if there’s a good probability (higher than 5%) that they were simply by chance.

And so this brief activity in about 20 minutes can help students develop what I think is the most important thing for them to take away from this part of the IA: the relationship between statistical significance and inferential statistical tests.