I pulled one of my favourite lessons this week for Year 7 and Year 9 – The Zener card experiment. If you are not aware of Zener cards, they are set of five symbols drawn on cards that are used in a classic ESP experiment.It’s the test used by Dr Venkman at the start of the 1984 historical documentary Ghostbusters. Essentially, one person (the experimentor) selects one of the cards at unseen at random and the other person (experimentee as I shall call them, because why not) attempts to use psychic powers in order to guess what is on the card. The reason I like this lesson in particular is that it gives some really nice insights into the problems surrounding experimental science.
I start by explaining the premise and introduce the hypothesis as an idea that can be tested. It’s usually something along the lines of ‘people have psychic powers’. At this point students like to vocalise their ideas, which is a nice way to demonstrate how bias can leak into any experiment when we are come expecting a certain result before the experiment has started. Not that it is necessarily avoidable, but it’s the awareness of the situation I am trying to develop. Next I give the idea of testing the hypothesis by asking a student to think of a capital city and claiming I can predict their answer. Of course, all I write on a piece of paper is “CORRECT” which I hold up triumphantly to whatever city they come up with. The obviousness of such a scam is a nice way to introduce the idea that people may not always be truthful when it comes to presenting results, and we should be prepared to question methodology and point out errors.
This is when things get more interesting. By doing a version of the experiment with just five cards (the real experiment used a deck of 25) it’s possible to show lots of examples of sensory leakage (for example seeing through the paper) or letting people glance the card before it’s shown. Because they want to get the answer ‘right’, the tendency for the students to try to cheat is remarkably high. Still, they get the idea of having to be secure in your method to gain valid results. The students then do repeats of 10 and record the number of correct responses. A class set of data can then be gathered (frequency graph optional) and the results discussed. Introducing random possibility at this point is a great starting point for statistical probability tests. Students can usually work out that there is a one in five chance of being right just by chance, so in a set of ten cards getting at least 2 cards right would actually be expected simply by random choosing. It makes more sense to students when they see this from the perspective of an ‘obvious’ experiment that results could randomly occur then when they are trying to impose the idea of stats test on an experiment where they may not understand large parts of the essential content.
Higher ability students can also be introduced to the concept of a null hypothesis when they realise that the experiment cannot exclude the possibility that someone might be psychic in a ‘different way’, taking them towards the idea that science works in probabilities and confidence rather than absolutes, which is of course only for the Sith. Interestingly, I always seem to get the highest frequency around 4/10, which can lead to interesting discussions about where the value for ‘significance’ lies; the basics of p values.
Writing up the experiment consists of saying of the data supported the conclusion, methodical flaws and improvements. As a way to study experimental design errors it is easily understood by students so it seems so “common sense” and they quickly spot outliers and question those students (in one way or another…) and get to see the importance of repeating data to eliminate outliers (high scoring individuals quickly tend to regress to the mean value). You can also look on YouTube for some videos of the cards to show a sequence with the human element removed as a source of bias. There’s a lot of that can be gained from an apparently simple test that can be applied to real-life science (cherry picking of data and confirmation bias come out strongly for example). The reason it works as an introduction to scientific methodology is that it removes the ‘science experiment’ part – there is no need for students to use up valuable thinking space on conceptualising abstract ideas like concentrations, enzymes, or absorption. The concept of this experiment is so straightforward that time can be spent analysing the background scientific understanding of experimentation. And finally you get to tell them at the end that no one has ever been able to demonstrate any kind of psychic abilities under controlled conditions. But who am I to deny anyone the chance to win Randi’s $1 million?