Like what you saw?
Create FREE Account and:
- Watch all FREE content in 21 subjects(388 videos for 23 hours)
- FREE advice on how to get better grades at school from an expert
- Attend and watch FREE live webinar on useful topics
M.Ed., Stanford University
Winner of multiple teaching awards
Patrick has been teaching AP Biology for 14 years and is the winner of multiple teaching awards.
Experimentation is the fourth step of the scientific method. Science experiments are necessary to test hypotheses and to form accurate conclusions based on real world results. All branches of science conduct experiments.
Science is all about testing hypotheses and in order to see whether or not they're right or wrong. No matter what you do when you design an experiment, it doesn't matter if you're using the most expensive equipment in the world if you don't have a good design. The key elements of a good experiment is one, you need to have a control group. Two, you need to have a large sample size and three you need to do blind experiments if you're working at all with people.
Let me go to this first thing about, what does it mean to have a control group. I may, I may suggest that perhaps I'm going to test it in a hypothesis that the best way to get from San Jose to San Francisco is to drive by Highway101 as opposed to driving say by City Streets or Highway 280. Well, in order for me to do this I need to have a comparison. So let's suppose I will say well, Highway 101, it's a highway, city streets they're not highway 280 that's also a highway. So one time I'll drive Highway 101 and the other time I'll drive 280, 280 being my control group that I use to compare my results to.
Now, you need to be careful that when you do this that you design your experiment such that you only have one variable, you don't have a bunch of them. What does that mean? Well, let's suppose I drive Highway 101 at say, 6:00 a.m. in the morning while I drive 280 at 9:00 a.m. in the morning. You may realize, hey, 6:00 a.m. there's hardly anybody on the streets while at n9:00 a.m. that's in the middle of rush hour. And that's added another dimension, that's added another variable called time instead of just which route I'm taking. So what I should do is, I should drive Highway 101 at say 9:00 a.m. and Highway 280 also at 9:00 a.m.
Now, this leads into the next thing about a large sample size. You may think that's kind of difficult. How am I going to drive both of them at the same time? Well, that's why I would need to drive it multiple times. By driving it several times I also get to avoid doing, running into some problems. One time when I was driving on the freeway, I saw this car about a hundred feet in front of me start to veer off a little bit to the right and I thought that's weird. He suddenly hit the shoulder, went like this and whipped around. He whipped so fast he flipped and actually it was rather impressive but he flipped several times, wind up in the middle of the highway on fire. That kind of slowed things down just a little bit. So by having a large sample size, by say you driving Highway 101 say 2 weeks in a row, I can eliminate the random affects that I can have. One day be slower or faster than the other. And then I drive Highway 280, say 2 weeks and again I'll average the results.
Now, some of you may be saying, "Hey, why not just have your friend drive 280 while I drive 101." Well that goes back to what your control group is. Remember it's only different by one variable and if I just have me versus one other person. Maybe my friend is a very cautious legal driver while I live out my race car fantasies in my mini van in 95 miles per hour. So that may influence it.
Now, you can sometimes get around this by having a large group. Say have 20 people drive 101 and 20 people drive 280. Not all on a bus, in separate cars. The best experiment of course would be one where you have a large group of people drive Highway 101 for a couple of weeks and a large group of people drive Highway 280 again for a couple of weeks and then you average the results.
Now, what do I mean by a blind experiment. Whenever you're working with people you need to have what's called a blind experiment. And that doesn't mean you start poking them in the eyes with sharp pointy sticks. That instead means you have to avoid them knowing, are they in the control group or your experimental group. Why is that? Well, your brain controls your body but your mind controls your brain. And this can cause some really bizarre results. Scientists found out sometime ago that you could give candy to somebody but if you convince them that it's actually medicine, their mind will get their brain to keep their immune system into a high gear and they'll actually get better. This is called the Placebo affect. This was well demonstrated to me by some of my AP biology students several years ago. They did a research project where they experimented on a bunch of Stamford students. And what they told the students is that they're looking at the effects that alcohol had on their motor skills and mental mood and such like that. What the Stamford students didn't realize is that they were being given non-alcoholic beer. Now my students had a couple of kids who, a couple of Stamford students who were their friends that they let it on on the joke and so they acted drunk. Everybody wound up feeling drunk and in fact one of the kids was caught on camera saying, this is like the fastest bus I've ever gotten. That explains why you really need to be careful about the Placebo effect.
Now, sometimes you also need to do what's called a double blind experiment. That's where the researcher also doesn't know if one person is in the control or variable group until they're finished doing all the analysis of the data. Now why would you want to do that, you ask. Well, imagine you're a cancer researcher and you've worked for 10 years to come up with a cure for cancer. Now you're finally doing the experiment, you're getting your results. If you know who is getting your cure versus who is not, you may accidentally [IB] not so accidentally, influence or bias your analysis or bias the results because if you're right you make a ton of money, if you're wrong you don't.
Please enter your name.
Are you sure you want to delete this comment?