luulapants shared this interesting story from middle school about an experiment they did in biology class. The teacher tried to throw them a curveball, but due to the wonders of a public school education everyone failed. Here is the full story:
Source: luulapants
(via: reddit)
I do this on purpose with an observational unit on the moon (5th grade). Kids think they know the moon, but they don’t. They just need to go out once each 24 hours and draw a picture of the moon as they see it in the sky. So they go out at night, where they expect the moon to be. Most come back and say they couldn’t see it, and conjecture that it must have already set, or it’s too low, or something is blocking it, like a tree or an apartment building. Some, however, come back with delightful drawings of full moons or crescents, which they DID NOT SEE. I know they didn’t, because we start at a the New Moon phase, and NOBODY can see the moon. This allows us to have a conversation about falsification of data, and how it’s more important to relate your actual observations, rather than what you think you should be seeing, or what you think the teacher wants to see. We also talk about how scientific consensus works. When you have 30 kids saying they saw one thing, and 2 kids saying another thing, you verify the data and move on from the kids who are sulking about being called out. We don’t give time to people who say they saw the crescent moon rise at sunset. This is, by the way, in a PUBLIC SCHOOL. Please people, stopping pissing on public schools because your own personal teacher was a bell-end.
Sure, your students may have walked away from the lesson with an important insight into data gathering. But if 99 other teachers fail to grasp what you did, most students will emerge from the schooling system with the mindset described in the post. And if your students take your lesson to heart and prioritize facts over the expectations of their bosses/contractors/clients, they will be punished for it. It IS a failure of the public school system if what you did is the exception as opposed to the rule.
At university in biochemistry class in one practical my results were way off what they should be. I consulted with a number of my class mates and found that some had gotten the predicted results and some (about a fifth) got similarly way off results. I collated all the sets of results (essentially a proxy for rerunning the experiment), reported the results that were as expected and explained why they were as expected then the anomalous results and hypothesised that some unaccounted for factor was to blame. I even identified that those people with anomalous results tended to have worked at one end of the lab and those who got the expected results at the other end but that I did not have any information on why that could have had an impact on the results. I got an F and a note that I must have done the experiment wrong and should have just got the ’right’ results of someone else (which I did, my report actually included results from over 80% of the class).
A few weeks later I’m chatting with one of the lab techs and they happen to mention how they wished different lecturers would co-ordinate their labs better so you wouldn’t have two year groups needing the same equipment at the same time. When we had our lab, which included incubating the experimental materials in a water bath at 80 degrees C, the year below us were doing a practical that also used water baths. As a result, the lab techs had had to get some older water baths out of storage and use them. The ones closest to the door of our lab were the old ones, the ones at the opposite end were the newer ones. Guess which end of the lab the people with anomalous results were at, older or newer water baths? Had I known this before writing my report I would have advanced the hypothesis that perhaps one or more of the older water baths was not at the correct temperature.
Here’s one from actual personal experience – this is what happened to me and my initial Doctoral thesis project (keeping it vague to not call out anyone on what happened):
So I was starting my research for my Doctorate in the sciences. Fortunately for me, I was given/had chosen a project that someone else had been working on, and I was going to further develop the concepts to see how it could be applied to more general systems and compare to what happened in real-life experiments. (This was all theoretical work, and the calculations were complicated for the computer simulations.) I took a while to figure out the theories that were used to create the equations, and then had to take some time to figure out how the program that the previous student. In the time that I was doing that, the previous student was wrapping up his research and successfully defended his thesis and graduated with his Ph.D. for the work he had done, so I had limited access to his time and had to do the background research on my own to catch up. After he was gone, and after I understood what I was working with, it was finally time to use his program to recreate his data as a base point. Here’s where it gets interesting…
You see, after running his program with the starting points that he had used in his thesis, I was SUPPOSED to see the data converging on a single value. What I was ACTUALLY seeing was the data wildly diverging to no single stopping point. Obviously my first reaction was that I was doing something wrong, so I reran the program multiple times, getting the same result every time (not an identical graph every time, but the same general chaos in the final results). I showed my data to my professor, and he was stumped about what could be happening. I went back to the theories and equations that were being used, and everything looked right there. But how was it that I could not use the same program with the same starting point and reach the same result as the (now graduated with his Ph.D.) previous researcher? That’s when I had an extremely troubling thought – what if the problem was not with the science or math, but with the program itself?
Now I had not created the computer program and the algorithm for doing the calculations, so I had to do more research about how it was supposed to work and how it was implemented. This took some more time in my project, but I learned about how the code was supposed to take small steps and find a minimum energy level bit by bit. Think of it like you are somewhere on the side of a mountain, and you can choose a random step each time in one of a few directions, and when you find the step that would put you downhill by the most amount, then you choose that step and repeat the process. Eventually you make enough steps that you should find yourself at the base of the mountain, which would be the “single value” you are looking for. This all makes sense, with one obvious assumption necessary to make it all work. WHAT IF YOU ARE NOT ON A MOUNTAIN, BUT INSTEAD ON A FLAT PLAIN? Sure, you can take multiple steps, and maybe there’s some small hill or bump in the ground that might make a particular step look better than others, but all in all you are just going to wander around until you end up at some random location.
What’s worse, since we are talking about computers, there is this thing called “floating point error” that makes things worse. Imaging you are calculating the number “one tenth” into the computer. You and I know that this will be 0.1 exactly, but in a computer this can end up being 0.10000000000001 or 0.099999999999 because it can not actually be 100% accurate all the time. To see this in action, take a calculator and divide 1 by 7, hit the equals key for the answer, then multiply that answer by 7. Now most people would think that one seventh multiplied by seven will just be 1, but the calculator will be a little off. (My phone gives 1.0000000003 as the result.) Now this is pretty close, but if you take this answer and do it again and again then this can lead to the either the same answer (if you’re lucky) or a wildly different result over time (if the randomness gets compounded each time). (Your calculator will probably give the same result each time, but a computer might not because it’s more accurate.) This was what I was seeing, and I presented my findings to the group. I determined that the calculations were not actually possible given the methods used, and that THE PREVIOUS STUDENT COULD NOT HAVE OBTAINED THE DATA THAT HE PRESENTED IN HIS THESIS. Yes, you read that right, he literally falsified his results so that he could graduate with his doctorate!
After the meeting, I talked with my professor and asked what I should do next, since I had just wasted a year of my academic career on falsified results. Needless to say, I had to choose another topic for my research, and the previous research was abandoned.