The real gremlin that wreaks havoc on an experiment is an extraneous variable that changes in a systematic way along with the independent variable. I a well – controlled experiment, the variation in the independent variable must be the only systematic variation that occurs across treatment conditions. If an extraneous variable occurs in one treatment condition but not another, an experiment cannot be internally valid. If uncontrolled extraneous variables are allowed to change along with the independent variable, we might not be able to tell whether changes in the dependent variable were caused by changes in the independent variable or by extraneous variable that also changes value across conditions.
When the value of an extraneous variable changes systematically across different conditions of experiment, we have a situation know as confounding. Box 6 – 3 presents a hypothetical experiment in which confounding is a serious problem. When ther is confurding, experimental results cannot be interpreted with certainty. Causal relationships between the independent variables cannot be interred. In effect, confounding sabotages the experiment because the effects we see can be explained equally well by changes in the extraneous variable or in the independent variable. Our experiment is not internally valid. In subsequent chapters we will study some of the basic techniques used to avoid confounding. Our goal is always to set up experiment in such a way that the independent variable is the only variable (besides the dependent variable) that changer value across conditions. To draw causal inferences about the effects of the independent variable, we must be sure that no extraneous variables change along with the independent variable.
What’s wrong with this experiment? Suppose a researcher was interested in the effects of age on communicator persuasiveness. She hypothesized that older communicators would be more persuasive than younger communicators – even if both presented the same argument. She set up an experiment with two experimental groups. Subject listened to either an 18 year old man or a 35 year old man presenting the same – minute argument in famor of gun control. After listening.
To one of the communicators, subjects rated how persuaded they were by the argument they had just heard. AS the researcher predicted, subjects who heard the older man speak were more persuaded. Would she be justified in stating that a speaker’s age influences persuasiveness? Hopefully, you identified this as a confounded experiment. Too many extraneous variables could have changed along with the independent variable (age): the older speaker mat have seemed more attractive, better educated, more intelligent, or more selfconfident (all these variables can influence our persuasiveness). Even though the researcher believed she was manipulating only the age of the speaker, several other extraneous variables might also have systematically varied along with the IV. Clearly, we could not say with assurance that age – rather than one of the other variables – influenced persuasion. What might she have done to control for these potential sources of confounding?
Classic Threats to Internal Validity
Psychologist Donald Campbell (Campbell, 1957; Campbell & Stanley, 1966 ; Cook & Campbell, 1979) identified eight kind of extraneous variables that can threaten the internal validity of experiments and quasi - experiments. Since that time, Campbell’s listing of these potential sources of confounding has become required course material for successive classes of experimental psychology students; hence, they have become known as the classic threats to internal validity.
The real gremlin that wreaks havoc on an experiment is an extraneous variable that changes in a systematic way along with the independent variable. I a well – controlled experiment, the variation in the independent variable must be the only systematic variation that occurs across treatment conditions. If an extraneous variable occurs in one treatment condition but not another, an experiment cannot be internally valid. If uncontrolled extraneous variables are allowed to change along with the independent variable, we might not be able to tell whether changes in the dependent variable were caused by changes in the independent variable or by extraneous variable that also changes value across conditions. When the value of an extraneous variable changes systematically across different conditions of experiment, we have a situation know as confounding. Box 6 – 3 presents a hypothetical experiment in which confounding is a serious problem. When ther is confurding, experimental results cannot be interpreted with certainty. Causal relationships between the independent variables cannot be interred. In effect, confounding sabotages the experiment because the effects we see can be explained equally well by changes in the extraneous variable or in the independent variable. Our experiment is not internally valid. In subsequent chapters we will study some of the basic techniques used to avoid confounding. Our goal is always to set up experiment in such a way that the independent variable is the only variable (besides the dependent variable) that changer value across conditions. To draw causal inferences about the effects of the independent variable, we must be sure that no extraneous variables change along with the independent variable. What’s wrong with this experiment? Suppose a researcher was interested in the effects of age on communicator persuasiveness. She hypothesized that older communicators would be more persuasive than younger communicators – even if both presented the same argument. She set up an experiment with two experimental groups. Subject listened to either an 18 year old man or a 35 year old man presenting the same – minute argument in famor of gun control. After listening. To one of the communicators, subjects rated how persuaded they were by the argument they had just heard. AS the researcher predicted, subjects who heard the older man speak were more persuaded. Would she be justified in stating that a speaker’s age influences persuasiveness? Hopefully, you identified this as a confounded experiment. Too many extraneous variables could have changed along with the independent variable (age): the older speaker mat have seemed more attractive, better educated, more intelligent, or more selfconfident (all these variables can influence our persuasiveness). Even though the researcher believed she was manipulating only the age of the speaker, several other extraneous variables might also have systematically varied along with the IV. Clearly, we could not say with assurance that age – rather than one of the other variables – influenced persuasion. What might she have done to control for these potential sources of confounding?
Classic Threats to Internal Validity
Psychologist Donald Campbell (Campbell, 1957; Campbell & Stanley, 1966 ; Cook & Campbell, 1979) identified eight kind of extraneous variables that can threaten the internal validity of experiments and quasi - experiments. Since that time, Campbell’s listing of these potential sources of confounding has become required course material for successive classes of experimental psychology students; hence, they have become known as the classic threats to internal validity.
การแปล กรุณารอสักครู่..
