Till somebody loses an eye.

When I worked in TCD, I helped run the Transition Year work experience weeks, which were held twice a year in my then Department. We used to ring-fence 10% of the places on the course for The Dispossessed: students from a school in TAP, the Trinity Access Programme, which privileged certain ghetto non-fee-paying schools with which we shared the Inner City. In general, kids from those schools wouldn't apply for the course [a response that goes with the territory of being dispossessed]; so I made some effort to push the option at the science teachers at each school. One of them was so taken with my enthusiasm that she asked if I'd be willing to guest-lecture a class on 'Science' with her 11 year olds.  That was quite a good fit because I wasn't limited to my published research but could do whatever!  Spoiled for choice, I thought I'd help them to generate some real data and then show them that they should be real careful with accepting such data - and indeed any data without critically evaluating it.

I generated a bunch of random numbers between 1 and 100 and had these ready on cards - one for everyone in the audience.  Each student was asked to reflect on and remember their number. I then asked them to estimate the population of Spain in millions given that Ireland is home to 4.6 million people, and to write this down on the back of the card.  The theory, according to the intrinsic bias concept of 'anchoring', was that students with high numbers would pitch their estimate of # Spaniards upwards while students with low numbers would guess lower.  I picked Spain because at 46.4 million [2015], the real number was about midway between 1 and 100. Well, that was a total bust. There was no relationship between card-numbers and population-numbers as was obvious when I plotted them out on the board. Maybe the sample size [N=30] was too small; maybe they didn't know what Spain was; but maybe the known fact of anchoring is a lot of nonsense.

I still had half a lesson to fill and I said that we would now carry out a hypothesis-driven experiment with dry ice = frozen carbon-dioxide. I had brought a big styrofoam box full of the stuff and also some 35mm film-canisters. I wrote before about how these handy little snap-lid plastic vials can be filled with dry ice and blow the lid off with a satisfying >!pop!< it's a handy party trick.  But that's not science.  Having an idea about some natural phenomenon and carrying out an experiment to better understand the natural world - that's science. For a science experiment you should have a hypothesis / theory to test; which means you have to focus on particular elements of the general phenomenon.  I showed them the party-trick from the teachers desk. Tghe taecher herself was seated at the back of the class to protect them if I went Postal or protect me if they rained me with spit-balls.

I showed the class me putting a small chunk of dry ice in the first canister and carried on talking about what a hypothesis was until the little lid spun up into the air with a satisfying crack. I suggested that we could do an experiment with black canisters and white ones to see if one set fired off before the other. I talked a bit about black-bodies and how they might absorb more heat from the room, that this might speed up the sublimation [solid -> vapor] process inside, and . . . so we had a testable hypothesis. "I have here 6 black and 6 white canisters; who wants to take part?; you have to time how long it takes between me adding the fragment of dry-ice and when the lid pops off; we can do this several times to see how reliable the experiment is".  Cascade of offers to mind one of the canisters.

I doled out the canisters and put in the pea of dry-ice and went back to the front of the room to await events. Then I noticed that a young chap with a buzz-cut had picked up his canister and was staring expectantly at the lip which was pointed directly at his eye. Nooooooo! I roared, put the canister down. Everyone put their canisters on their desk and sit back out of the way. Pop! . . .  pop . . . pop-pop-pop-pop-pop. This is what it looks like almost too fast to see . . . like a bullet. Nobody was injured but my hysterical shout dropped my street-cred through the floor. The experiment as stated didn't work very well either: the data were noisy as well as loud and this hypothesis was neither accepted nor rejected with that sample size. Out the corner of my eye I could see the teacher appreciate just how far I was adrift from what she thought was advisable to carry out with her students.