# How to launch an experiment.

Is throwing paper airplanes against the rules at your school? Here's a way around the problem: an airplane-flying activity your teacher will love. By doing it, you'll learn the important parts of a good science experiments. Start by following the procedure on the next page.

PROCEDURE

1. Fold paper airplane as shown above.

2. From a "starting line," launch your airplane. Observe its flight path.

3. Record your observations in the data table on page 10.

4. Now change the plane slightly by flipping the tail up.

5. launch the changed plane and observe its flight path. Record your observations.

CH-CH-CH-CHANGES

Whether you know it or not, you just performed an experiment. In an experiment, you change something--a variable--to see what happens. Here, you changed the design of the paper airplane by flipping the tail up. You made this change on purpose, to learn about its effect.

Variables you change on purpose are called independent variables. When you change an independent variable--such as the design of the paper airplane--you are looking for an effect, or response, in another variable. In this case, you wanted to know how the flight path changed. This responding variable is called a dependent variable.

Any good experiment should have only one independent variable. Why? Because if you change more than one independent variable at the same time--say, by changing the design of the plane and the paper it is made of--you'll never know which change caused the response in the dependent variable, or how much each change contributed to the response. So you have to test each one separately.

Once you've selected which independent variable you'll test, make sure all other potential variables remain constant (the same) during your experiment. What are some things you should keep constant each time you try this experiment?

UNDER CONTROL

If you only wanted to know if the tail-up plane will fly more smoothly, couldn't you make just that one and test it?

Think about it. How would you know what effect tail position has without something to compare it to: a control. Without a control, or standard for comparison, your tail-up results (like the results from any experiment) would be meaningless.

Sometimes controls are easy to spot. Say you want to study the effect of light on plant growth. An obvious control would be to try to grow the plants without any light.

Now say you want to study how different amounts of light (or colors, or kinds) affect plant growth. Could you still use the same control? Sure. You're not changing the independent variable (light). You're just changing something about it. For lack of a better word, we call that changing the level of the independent variable. The "no light" level is still a good control.

In some experiments, however, it is impossible to have a control that receives no "treatment." For example, if you're studing the effect of plane design on flight pattern, you can't compare one design to "no" design. Every plane has some kind of design. So how do you do your experiment? You pick one design to be your control and compare all the others to it.

REPEAT, REPEAT . . .

Once you know your experiment has all the right parts, how do you make sure your results weren't a fluke--caused by a sudden wind or some other chance event? The answer is simple: do repeated trials. Test each level of your independent variable several times, recording all results. If you get similar results each time, you can be more confident that your findings were not caused by error or chance.