In our experiment, we will coat sheets of glass with different SPF ratings of sunscreen lotion and put them under bright sunlight. We will then measure the UV radiation behind the coated glass.

To make sure our experiment is accurate, we will set up several identical experiments (called "controls") with only the SPF rating changing. We will keep all the other variables the same for every experiment, such as the brand of sunscreen lotion, the amount of sunscreen lotion used, the type of materials and the type of UV meter, and the temperature, lighting conditions, and humidity where we do the experiment. We need to do the experiment when the sun is shining and there are no clouds. We will also make sure to buy all the sunscreen lotion from the same pharmacy and at the same time, in case different batches of the same brand have different specifications. We will use the same UV meter for each control experiment, because different meters (even if they are the same brand and model) can give different results for different reasons (like one UV meter might have run out of batteries but the other one didn't).

All of these variables are called control variables. When you design your experiment, you need to include steps to measure the results of all the experiments.

You need to decide on the independent variable and how to measure how it affects the dependent variable. Remember, there can only be one independent variable. All the other variables have to stay the same. This is so we can be sure that any changes in the dependent variable are because of the independent variable. In our example, we need to see how changing the SPF rating affects the level of protection. If we don't keep everything else the same except for the SPF level (like if one control experiment is done in bright sunlight and another is done at sunset), we won't know if the changes we see in the results are because of the SPF level. Your procedure should also say how many trials to do. You should do your experiment at least three times, but you can do it more if you want. For experiments that involve surveys, you don't need to ask the same person the same question more than once. But for this kind of experiment, you should make sure to survey a lot of people for more reliable results.