Does RealClean detergent make your clothes cleaner? The smiling company representative in the television commercial takes two identical shirts, pours something messy on each one, and drops them into identical washing machines. RealClean brand detergent goes into one machine and the recommended amount of a rival brand into the other. Each washing machine is set to the same cycle, for the same period of time, and the ad fast-forwards to show the continuously smiling representative taking the two shirts out. Guess which one is cleaner.
Now, it would be very easy to rig the demonstration so that RealClean does a better job: the representative could use less of the other detergent, use an inferior-performing washing machine, put the RealClean shirt on a soak cycle forty-five minutes longer than for the other brand, employ different temperatures, wash the competitor's shirt on the delicate rather than regular cycle—I'm sure you can think of a lot of ways that RealClean's manufacturer could ensure that its product comes out ahead. It would be a bad sales technique, however, because we're familiar with the direct experimental type of test, and someone would very quickly call, "Foul!" To convince you that they have a better product, the makers of the commercial have to remove every factor that might possibly explain why the shirt came out cleaner when washed in their product. They have to hold constant or control all these other factors—type of machine, length of cycle, temperature of the water, and so on—so that the only reasonable explanation for the cleaner shirt is that RealClean is a better product. The experimental method— performed fairly—is a very good way to persuade people that your explanation is correct. In science, too, someone will call, "Foul!" (or at least, "You blew it!") if a test doesn't consider other relevant factors.
Direct experimentation is a very powerful—as well as familiar—research design. As a result, some people think that this is the only way that science works. Actually, what matters in science is that explanations be tested, and direct experimentation is only one kind of testing. The key element to testing an explanation is to hold variables constant, and one can hold variables constant in many ways other than being able to directly manipulate them (as one can manipulate water temperature in a washing machine). In fact, the more complicated the science, the less likely an experimenter is to use direct experimentation.
In some tests, variables are controlled statistically; in others, especially in biological field research or in social sciences, one can find circumstances in which important variables are controlled by the nature of the experimental situation itself. These observational research designs are another type of direct experimentation.
Noticing that male guppies are brightly colored and smaller than the drab females, you might wonder whether having bright colors makes male guppies easier prey. How would you test this idea? If conditions allowed, you might be able to perform a direct experiment by moving brightly colored guppies to a high-predation environment and monitoring them over several generations to see how they do. If not, though, you could still perform an observational experiment by looking for natural populations of the same or related species of guppies in environments where predation was high and in other environments where predation was low. You would also want to pick environments where the amount of food was roughly the same—can you explain why? What other environmental factors would you want to hold constant at both sites?
when you find guppy habitats that naturally vary only in the amount of predation and not in other ways, then you're ready to compare the brightness of color in the males. Does the color of male guppies differ in the two environments? If males were less brightly colored in environments with high predation, this would support the idea that brighter guppy color makes males easier prey. (What if in the two kinds of environments, male guppy color is the same?)
Indirect experimentation is used for scientific problems where the phenomena being studied—unlike color in guppies—cannot be directly observed.
Was this article helpful?