Designing Experiments Using the Scientific Method

>

Figure 1: Graph showing the time each mile of a marathon was reached for a runner who consumed pasta (white dotted line), a runner who consumed coffee (squared line), and a runner who slept four extra hours prior to the race (black dotted line).

Taking control of your experiment

How would you know if your race times were improved either by eating pasta or drinking coffee? You would have to run a marathon without eating pasta the night before or drinking coffee the morning of the race. (Exhausted yet?) This marathon would be your control. A control is a set of base values against which you compare the data from your experiments. Otherwise, you would have no idea if your results were better, worse, or the same.

Drawing conclusions

So, maybe it took you less time to reach each mile along the marathon route after the night of pasta eating, but your race times after drinking the coffee matched those of the control. That would support your initial hypothesis, but it would refute your second hypothesis. There's nothing wrong with being wrong, as long as the information is useful. Knowing what doesn't work is just as important as knowing what does.

Your conclusion to these two experiments would be something like: "Consuming pasta the night before a 26-mile marathon improves race time, but consuming caffeine has no effect."

However, in scientific experiments you have to confess your mistakes. This confession lets other scientists know what could be affecting your results. Then, if they choose to repeat the experiment, they can correct for those mistakes and provide additional beneficial information to the knowledge base. In the pasta-caffeine-race experiment, if you had consumed the pasta the night before and then the caffeine the morning of the race, your major error would be that of including more than one independent variable.

Another error would be having too small of a sample. A more accurate determination could be made by recording the race times at each mile for many runners under the same conditions (i.e., having them eat the same amount of pasta the night before a race or consuming the same amount of caffeine the morning of a race). Of course, their individual control times without those variables would have to be taken into account. Science. It's all in the details.

>
dummies

Source:http://www.dummies.com/how-to/content/designing-experiments-using-the-scientific-method.html