Experimental research in psychology applies the scientific method to achieve the four goals of psychology: describing, explaining, predicting, and controlling behavior and mental processes. A psychologist can use experimental research to test a specific hypothesis by measuring and manipulating variables. By creating a controlled environment, researchers can test the effects of an independent variable on a dependent variable or variables.
For example, a psychologist may be interested in the impact of video game violence on children's aggression. The psychologist randomly assigns some children to play a violent video game for 1 hour and other children to play a non-violent video game for 1 hour. Then the psychologist observes the children socialize afterwards to determine if the children in the "violent video game" condition behave more aggressively than the children in the "non-violent video game" condition. In this example, the independent variable is video game group. Our independent variable has two levels: violent video games and non-violent video games. The dependent variable is the thing that we want to measure—in this case, aggressive behavior.
Independent and Dependent Variables
In an experimental study, the independent variable is the factor that the experimenter controls and manipulates. This variable is hypothesized to be the cause of a particular outcome of interest. The dependent variable, on the other hand, depends on the independent variable, and will change (or not) because of the independent variable. The dependent variable is the variable that we want to measure (as opposed to manipulate). In a simple experiment, a researcher might hypothesize that cookies will make individuals complete a task quicker. In one condition, participants will be offered cookies if they complete a task, while in another condition they will not be offered cookies. In this case the presence of a reward (receiving cookies or not) is the independent variable, and the time taken to complete the task is the dependent variable.
Screenshot 2015-07-26 11.46.24.png
Effects of receiving a cookie as a reward (independent variable) on time taken to complete task (dependent variable). As shown in the figure, participants who received a cookie took much less time to complete the task than participants who did not receive a cookie.
An experiment can have more than one independent variable. A researcher might decide to test the hypothesis that cookies will make individuals work harder only if the task is easy to begin with. In this case, both the presence of a reward and the difficulty of the task would be independent variables.
Experimental Design
The purpose of an experiment is to investigate the relationship between two variables to test a hypothesis. By using the scientific method , a psychologist can plan and design an experiment that will answer the research question. The basic steps of experimental design are:
- Identifying a question and performing preliminary research to determine what is already known
- Creating a hypothesis
- Identifying and defining the independent and dependent variables
- Determining how the independent variable will be manipulated and how the dependent variable will be measured
The Scientific Method
The scientific method is the process by which new scientific knowledge is gained and verified. First you must identify a question and, after some preliminary research, form a hypothesis to answer that question. After designing an experiment to test the hypothesis and collecting data from the experiment, a scientist will draw a conclusion. The conclusion will either support the hypothesis or refute it. The scientist will then either reformulate the hypothesis or build upon the original hypothesis. The scientific method cannot prove a hypothesis, only support or refute it.
Experimental Design: Important Principles
A poorly designed study will not produce reliable data. There are key components that must be included in every experiment: the inclusion of a comparison group (known as a "control group"), the use of random assignment, and efforts to eliminate bias. When a study is designed properly, the only difference between groups is the one made by the researcher.
Control Groups
Control groups are used to determine if the independent variable actually affects the dependent variable. The control group demonstrates what happens when the independent variable is not applied. The control group helps researchers balance the effects of being in an experiment with the effects of the independent variable. This helps to ensure that there are no random variables also influencing behavior. In an experiment monitoring productivity, for instance, it was hypothesized that additional lighting would increase productivity in factory workers. When workers were observed in additional lighting they were more productive, but only because they were being watched. If a control group was also observed with no additional lighting this effect would have been obvious.
Random Assignment
To minimize the chances that an unintended variable influences the results, subjects must be assigned randomly to different treatment groups. Random assignment is used to ensure that any preexisting differences among the subjects do not impact the experiment. By distributing differences randomly between the conditions, random assignment lowers the chances that factors like age, socioeconomic status, personality measures, and other individual variables will affect the overall group's response to the independent variable. Theoretically, the baseline of both the experimental and control groups will be the same before the experiment starts. Therefore, if there is a difference in the behavior of the two groups at the end of the experiment, the only reason would be the treatment given to the experimental group. In this way, an experiment can prove a cause-and-effect connection between the independent and dependent variables.
Blinding and Experimenter Bias
To preserve the integrity of the control group, both researcher(s) and subject(s) may be "blinded." If a researcher expects certain results from an experiment and accordingly unknowingly influences the subjects' responses, this is called demand bias. If the experimenter inadvertently interprets the information in a way that supports the hypothesis when other interpretations are possible, it is called the expectancy effect. To counteract experimenter bias, the subjects can be kept uninformed on the intentions of the experiment, which is called single blinding. If the people collecting the information and the participants are kept uninformed, then it is called a double blind experiment. By using blinding, a researcher can eliminate the chances that they are inadvertently influencing the outcome of the experiment.
Counterbalancing
When running an experiment, a researcher will want to pay close attention to their design to avoid error that can be introduced by not balancing the conditions properly. Consider the following example. You are running a study in which participants complete a task of pressing button A with their left hand if they see a green light and pressing button B with their right hand if they see a red light. You find support for your hypothesis that red stimuli are processed more quickly than green stimuli. However, an alternative explanation is that people are faster to respond with their right hand simply because most people are right-handed. The solution to this problem is to "counterbalance" your design. You will randomly assign 50% of your participants to respond to the red stimulus with their right hand (and green with their left) and assign the other 50% to respond to the red stimulus with their left hand (and green with their right). In this manner, you are anticipating and controlling for this extra source of error in your design.
Strengths and Weaknesses of Experimental Research
One of the main strengths of experimental research is that it can often determine a cause and effect relationship between two variables. By systematically manipulating and isolating the independent variable, the researcher can determine with confidence the independent variable's causal effect on the dependent variable. Another strength of experimental research is the ability to assign participants to different conditions through random assignment. Randomly assigning participants to conditions ensures that each participant is equally likely to be assigned to one condition or another, and that there are no differences between experimental groups.
Although experimental research can often answer the causality questions that are left unclear by correlational studies, this is not always the case. Sometimes experiments may not be possible or ethical. Consider the example of the studying the correlation between playing violent video games and aggressive behavior. It would be unethical to assign children to play lots of violent video games over a long period of time to see if it had an impact on their aggression. Additionally, because experimental research relies on controlled, artificial environments, it can at times be difficult to generalize to real world situations, depending on the experiment's design and sample size. If this is the case, the experiment is said to have poor external validity, meaning that the situation the participants were exposed to bears little resemblance to any real-life situation.