Experimental and Correlational Research Methods for AP Psychology (page 2)
Practice questions for this study guide can be found at:
Some psychologists conduct experimental research in laboratories designed for carefully controlling conditions and measuring behavior.
The Controlled Experiment
The laboratory is one of the places where scientists test hypotheses, predictions of how two or more factors are likely to be related. Variables are factors that can have different values. In a scientific experiment, the researcher systematically manipulates a variable under controlled conditions, and observes the response. The factor the researcher manipulates is called the independent variable (IV). The dependent variable (DV) is the behavior or mental process that is being measured, the factor that may change as a result of manipulation of the independent variable. If the dependent variable changes when only the independent variable is changed, the researcher can conclude that the change in the independent variable caused the change in the dependent variable. Thus, the independent variable is the cause, and the dependent variable is the effect. A controlled experiment is the only research method that can establish a cause and effect relationship.
An effective way to determine the independent and dependent variables is to word the hypothesis in the form of an "If… , then…" statement. What follows the "If " is the independent variable (cause), and what follows the "then" is the dependent variable (effect). For example, "If students study for a quiz before going to sleep, rather than in the morning, then they will get higher scores on the quiz." Studying for a quiz before going to sleep, rather than in the morning is the independent variable and cause. Getting a higher score on the quiz is the dependent variable and effect.
For example, an experimenter hypothesizes that sleeping after studying for a biology quiz in the evening is more effective than studying for the same amount of time after waking in the morning. The population includes all of the individuals in the group to which the study applies (all of the students enrolled in introductory biology courses at the university for this example). To save time and money, most researchers use a subgroup of the population called a sample in their experimental research. The larger the sample size, the more likely it is to represent the population. The sample must fairly represent the whole group. This is achieved when every member of the population has an equal chance of being selected for the sample, when participants are selected randomly. Random selection can be achieved by putting all of the names in a hat and picking out a specified number of names, by alphabetizing the roster of enrollees and choosing every fifth name, or by using a table of random numbers to choose participants. These are examples. To test the hypothesis, the scientist needs to randomly assign some subjects to an experimental group that receives the treatment and to randomly assign others to a control group that does not receive the treatment. The control group is a comparison group. This is called a between-subjects design because the participants in the experimental and control groups are different individuals. Everything is similar between the experimental group and the control group except for the independent variable. Random assignment of participants to the experimental and control groups minimizes the existence of preexisting differences between the two groups. Differences between the experimental group and the control group other than those resulting from the independent variable are called confounding variables.
Confounding variables limit confidence in research conclusions. All participants, also called subjects, attend the same two sessions upon which the quiz is based. The experimental group is permitted to study for the quiz for 1 hour in the evening before going to sleep while the control group watches an unrelated comedy show. The control group studies for the quiz for 1 hour in the morning after awakening. The experimental group watches the comedy show in the morning at the same time. Everyone eats breakfast together, then they all take the same quiz. If the experimental group scores significantly higher than the control group, the experimenter can say that the results support the hypothesis. How does the experimenter measure effectiveness of studying? The experimenter uses the score on the quiz as the operational definition of effectiveness of studying. An operational definition describes the specific procedure used to determine the presence of a variable.
In order to attribute a particular result to a specific factor, the controlled experiment must limit variables. Confounding variables that could contribute to the effect must be eliminated. Subjects in the biology quiz experiment need to share the same environmental factors; they need to eat the same foods, sleep in similar beds in the same rooms, sleep for the same amount of time, etc.
Eliminating Confounding Variables
Experimenter bias (also called the experimenter expectancy effect) is a phenomenon that occurs when a researcher's expectations or preferences about the outcome of a study influence the results obtained. This is a special kind of confounding variable that can occur when a researcher is unaware that he or she is treating either the experimental group or control group differently from the other. A simple smile when addressing the experimental group that is not also shown to the control group qualifies as experimenter bias and as a confounding variable. The clues participants discover about the purpose of the study, including rumors they hear about the study suggesting how they should respond, are called demand characteristics. To eliminate the effects of demand characteristics, experimenters use the single-blind procedure, a research design in which the participants don't know which treatment group—experimental or control—they are in. To eliminate the effects of both experimenter bias and demand characteristics, experimenters use the double-blind procedure, a research design in which neither the experimenter nor the participants know who is in the experimental group and who is in the control group. The double-blind is most easily accomplished when a second experimenter or assistant who doesn't know the hypothesis or group assignments administers the experiment, keeping the principal investigator away from the subjects. When a number of factors might be responsible for an observed effect, to determine which deserves the credit, an experimenter needs to systematically manipulate or vary one or more factors while holding constant all the others that might be important. The effects of these manipulated events on some behavioral reaction are then assessed. It is then possible to demonstrate whether one factor is responsible for the result or if an interacting package of factors is involved.
In experiments involving drugs, participants in the experimental group usually receive the drug with the active ingredient, while subjects in the control condition receive a drug that seems identical, but lacks the active ingredient. The imitation pill, injection, patch, or other treatment is called a placebo. Subjects sometimes believe that the treatment will be effective, and they think they experience an improvement in health or well being. This is the placebo effect. The placebo effect is now used to describe any cases when experimental participants change their behavior in the absence of any kind of experimental manipulation. The experiments need not involve drugs at all.
A research design that uses each participant as his or her own control is called a within-subjects design. For example, the behavior of an experimental participant before receiving treatment might be compared to his or her behavior after receiving treatment. Two treatments might be tried. If two treatments are used, the order of the treatments could cause an effect. To eliminate the possibility, psychologists use counterbalancing, a procedure that assigns half the subjects to one of the treatments first and the other half of the subjects to the other treatment first.
Quasi-experimental research designs are similar to controlled experiments, but participants are not randomly assigned. Experimental research designs to study differences in behavior between men and women, boys and girls, young and old, or students in one class and students in another class are "sort of " experiments or quasi experiments. Because of confounding variables—preexisting differences between the experimental group and comparison groups—quasi-experiments do not establish cause and effect relationships, although they can point in the direction of them.
Although experiments conducted under carefully controlled conditions help establish cause and effect relationships, the time, expense, and artificiality of the environment limits that type of research. Psychologists more often use descriptive and correlational research methods such as survey methods that involve interviews or questionnaires, tests, and naturalistic observation. Correlational methods look at the relationship between two variables without establishing cause and effect relationships. The goal is to determine to what extent one variable predicts the other. Many factors that seem to be causally related are not. Often it's a third factor that causes the other two.
Naturalistic observation is carried out in the field where naturally occurring behavior can be observed. Naturalistic observation studies gather descriptive information about typical behavior of people or other animals without manipulating any variables. For example, Jane Goodall's team of scientists has been observing the behavior of chimpanzees in the wild for decades. Such studies have enabled scientists to predict when the chimps will fight each other or when they will mate. Similarly, other scientists have been studying human behavior in the workplace, in schools, in bars, etc. The data can be used for correlational analysis or for generating ideas for other research.
In the survey method, researchers use questionnaires or interviews to ask a large number of people questions about their behaviors, thoughts, and attitudes. In order for the information to be useful, the participants in the study should be representative of a larger population, which can best be achieved by random sampling. Accuracy of data is an issue because people sometimes distort their answers to appear more "politically correct," or they fail to recall information correctly. The data from surveys can be used for correlational analysis or for generating ideas for other research.
Retrospective or ex post facto studies look at an effect and seek the cause. For example, when researchers found an increase in babies being born with deformed limbs, especially in England but also in the United States and other western countries in the early 1970s, they asked the mothers of the babies many questions, then compared the answers of all of the mothers through correlational analysis. They found the strongest relationship between the mother taking the drug thalidomide during the pregnancy and the appearance of the limb deformities in the babies. Controlled experiments with rodents verified that the drug caused abnormal limb development in the babies of the animals.
Tests are procedures used to measure attributes of individuals at a particular time and place. Like surveys, tests can be used to gather huge amounts of information relatively quickly and cheaply. Results of tests can be used for correlational analysis or for generating ideas for other research.
For surveys or tests to be accurate measures of behaviors or mental processes, they must be both reliable and valid. Reliability is consistency or repeatability. Subjects should answer questions the same way on two different occasions. A subject should also get the same score on a test on two different occasions. Validity is the extent to which an instrument measures or predicts what it is supposed to. Questions about frequency of showering would not be valid indicators of cooking ability. Algebra questions would not be valid measures of what you learned in this chapter.
Another research method, the case study method, is an in-depth examination of a specific group or single person that typically includes interviews, observations, and test scores. The intensive description and analysis of the small group or individual is especially useful for understanding complex or rare phenomena. For example, case studies done on patients with gunshot wounds to the head enabled scientists to better understand how the loss of brain tissue affected specific aspects of behavior. Case studies have enabled us to better understand a wide range of topics, from how the brain processes information to autism. Clinical psychologists frequently do case studies.