Experimentation

Definition

In its simplest form, experimentation is a method of determining the presence or absence of a causal relationship between two variables by systematically manipulating one variable (called the independent variable) and assessing its effect on another variable (called the dependent variable).

Importance and Consequences

The hallmark of experimentation is that it allows researchers to make statements about causality. There are several features of experiments that facilitate such claims:

• 1.
  Experiments allow researchers to create a situation in which changes in the independent variable precede assessment of the dependent variable—making it possible to draw conclusions about the directionality of the relationship. This is important, because to establish a cause-and-effect relationship between two events, the event that one supposes to be the cause must precede the event that one supposes to be the effect.
• 2.
  Experimentation entails randomly assigning participants to experimental groups. When random assignment is employed, each participant has an equal chance of being assigned to any of the conditions in a study. This technique allows researchers to assume that the experimental groups of participants are equivalent at the outset of the study. Thus, researchers can safely attribute any observed differences in the dependent variable to the experimental manipulation without worrying about the possibility that naturally occurring differences between the groups of participants could account for these differences.

For example, suppose that a researcher wants to determine whether playing chess causes an increase in creativity among fifth graders. Imagine that the researcher decides to have two groups: one group that plays chess for an hour after school each day for 6 months and a control group that has free time for an hour after school each day for 6 months. At the start of the study, the researcher recruits a group of 80 fifth graders and allows them to sign up to either play chess or get free time. Six months later, the researcher gives all the kids a creativity test. Sure enough, the researcher finds that the 12 kids who chose playing chess scored higher on the creativity test than the 68 kids who chose free time. Can the researcher say that chess caused an increase in creativity? No—because the kids who chose to play chess might have been more creative than those who chose free time. Thus, the group differences in creativity may have been there from the start and may have had nothing to do with the researcher's manipulation. To do this experiment properly, the researcher should randomly assign the students to a condition, so that 40 played chess and 40 had free time. By doing so, the researcher could assume that the two groups were equivalent at the start of the study, and so any differences in creativity at the end of the study could be attributed to the independent variable (in this case, playing chess vs. having free time) and not to differences in creativity that existed before the study began.

3. Experimentation allows researchers to isolate the effect of the independent variable by controlling all other elements of the environment, thereby ensuring that all of the participants in a given study undergo a similar experience, with the exception of the experimental manipulation. In the chess versus free time example, imagine that the kids in the chess group always listened to classical music while they played chess, whereas the kids in the free time group did not listen to music. At the end of the study, could the researcher be sure that the difference in creativity between the two groups was due to the game that they played? No—because whether the kids listened to classical music also may have influenced their creativity. To do the experiment properly, the two groups should be identical with the sole exception of the independent variable (chess vs. free time). In this way, the experimenter could be sure that it was really the independent variable that influenced the student's creativity and not some other factor.

...