多组实验

Choosing the correct experimental design is essential to answer the specific scientific question at hand.

In an experiment, researchers are concerned with variables—what changes. For instance, the researcher manipulates the independent variable to detect possible differences amongst participants.

The independent variable can have different levels, also known as conditions, which may result in different outcomes—what the researcher measures as the dependent variable.

If an independent variable has three or more conditions, the experiment consists of a multiple-group design. This is in contrast to a simple experimental design, which contains two levels, the experimental and control groups.

Each design is used to answer different questions; a two-group design tells you whether the independent variable has any effect, whereas a multiple-group design tells you how much of an effect each condition has.

Using a multi-group approach, this video demonstrates how to design levels of variables and conduct the study, as well as how to analyze data and interpret participants’ attitudes towards complex ethnic interactions.

In this multi-group experiment, participants are randomly assigned to one of four conditions: Colorblind, Multicultural, Control, or an Empty-Control group.

Two of the groups, Colorblind and Multicultural, are considered main experimental conditions, in which participants are unknowingly exposed to diverse perspectives through a word search task that focuses on shared similarities or appreciative differences.

In contrast, participants in the Control group are exposed to a mundane word search task that lacks any perspective, whereas participants in the Empty-Control group are not exposed to the task. This latter condition provides a baseline response of how participants act without any treatment.

For the word search task, participants are given words that relate to the attitudes of their assigned conditions. For example, colorblind words include: equality, unity, sameness, similarity, and blind; multicultural words include: culture, variety, difference, diversity, and multi; and control words include: practical, relaxed, logic, creativity, and friendship.

In addition, all of the word searches contain the same five distractor words: flowers, stars, artistic, world, and music. The distractors are included to prevent participants from guessing what the true nature of the study is—a concept referred to as hypothesis guessing.

Following the word search task, all participants are handed a photograph of someone they think will help them complete additional tasks when they return. In reality, the photograph represents a member of a social group with whom the participant does not identify—the outgroup member.

The dependent variable—the participants’ attitudes toward the outgroup member—is then quantified by how close they choose to sit near their supposed partner’s belongings. Note that the outgroup member never makes a physical appearance.

In this case, choosing to sit closer represents a more positive attitude towards the outgroup member versus sitting farther away.

It is hypothesized then that those who are exposed to the multicultural perspective will display more favorable attitudes toward an outgroup member when compared to those who are exposed to the colorblind perspective.

To conduct the study, create three different word searches that consist of five words associated with colorblind, multicultural, or mundane attitudes. Make sure that each search includes the same five distractor words.

Randomly organize the word searches inside packets to ensure that participants’ assignments are entirely based on chance and not any preconceived assumptions about the participant. Note that the packet for the Empty-Control group is indeed empty.

In an adjoining room, set up eight chairs. Place the outgroup member’s belongings on the far left chair, giving the participant seven seats to choose from.

To begin the experiment, meet the participant at the lab. Provide each participant with informed consent, which consists of a brief description of the research and procedures, and the potential risks and benefits of participating.

To assign the conditions, hand all participants a word search packet, except for those assigned to the Empty-Control group.

Once the word search is finished, escort the participant to an adjoining room. Hand each participant the same photograph of another person—the outgroup member—and tell them that they will now work together with this person on additional word search tasks upon their return.

Further explain to the participant that their partner has already arrived, but had to run out to their car to get something. Tell them to have a seat.

After participants choose their seat, walk to a different part of the room and note each one’s seat selection.

At the conclusion of the experiment, debrief participants by telling them the nature of the study.

Explain that deception was necessary to capture the participants’ natural reactions, as divulging the true intentions of the study beforehand could have influenced their behavior to meet perceived expectations.

To analyze how different perspectives influence participants’ attitudes, average the seat choice scores in each group and plot the means across the four conditions. To determine if group differences were found, perform an analysis of variance comparing the four groups.

Note that those in the multicultural group sat closer to the outgroup member, which translates to a higher attitude score. Thus, the exposure to the multicultural ideology led to the most favorable attitude as compared to all other conditions.

Now that you are familiar with multiple group designs, let’s take a look at how other researchers manipulate conditions to maximize experimental control.

For example, to determine the relationship between time spent sleeping and exercise performance, multiple groups are needed. That way, an optimal amount of sleep can be determined, where ultimately the longest amount of sleep may not help, but rather hurt, exercise performance.

Similarly, a multiple group design would be necessary to determine the optimal dose of a medication that achieves the desired outcome without side effects.

A multi-group design is also beneficial to test whether participation in a dance class relieves depression symptoms. In this case, adding a group exposed to social interaction without dancing controls for a potentially confounding variable—the inadvertent social interaction derived from the class.

You’ve just watched JoVE’s introduction to multi-group experimental design. Now you should understand how to design a multi-group experiment by including different levels of the independent variable.

Using a specific example, the video demonstrated how to conduct a multi-group experiment, as well as how to evaluate the results. Finally, through a discussion of applications, you should have a good understanding of how multi-group experimental design can be used to meet specific research needs.

Thanks for watching!