Two-Group Random Assignment Pretest–Posttest Design

Experimental Research and Internal Validity

Experimental research is a type of research that strives to explain causal relationships between variables. Although many forms of experimental design exist, many have common elements including measurement of both independent and dependent variables and the introduction of a treatment or manipulation. Often, these factors are combined with placement into different groups, which allows researchers to make comparisons across the groups and determine different effects of the treatment on each.

Researchers who seek to explain how independent variable A affects dependent variable B must carefully choose a design that enables them to rule out any other possible explanations of the changes observed in B. The assurance that no other possible causes exist is known as internal validity. This is particularly important in designs with more than one condition. For example, if a researcher wanted to study the effects of room color on mood, the rooms to which participants are exposed should be identically furnished, heated, lighted, and so on; otherwise, any changes in mood may be due to comfort, temperature, or lighting. By keeping all other elements identical, or constant, the researcher can be more certain that effects on the dependent variable (mood) are due only to changes in the independent variable (room color), thus increasing internal validity.

Another important way to increase internal validity in designs with multiple groups is random assignment, which is the process of assigning participants to treatment conditions without any possible bias. This is achieved through any process of random selection, such as flipping a coin and having those whose coin lands on heads be in one group and tails, the other. If participants are truly randomly selected, each individual has an equal chance of being in any condition. This ensures that any bias in the population (e.g., similar tendencies due to shared characteristics) does not impact the results of a study. For example, imagine a researcher studying the effects of music on test-taking anxiety. He or she creates two groups, one which is exposed to music while taking a test and another which is not, and compares the two to determine which group experiences more anxiety. If he or she assigns all female participants to one group and all male participants to the other, any difference in anxiety between the groups may be due to gender rather than exposure to music. By randomly assigning participants to groups, the researcher should have an equal number of each gender in each group and can be more certain that any difference between them is a direct result of only his or her manipulation. Random assignment works with other personality and physical differences, as well; ideally, any differences in participants (e.g., intelligent vs. unintelligent, right-handed vs. left-handed, or introverted versus extroverted) will be randomly distributed across conditions and create only random variation within groups.

...