touching them after handling diseased tissue was true not only for his hospital, but also for all other hospitals. Thus, in addressing the question of external validity of Semmelweis’s work, we would infer that his answers could be generalized to other hospitals with other women and not just to his own original setting. External validity, therefore, refers to the possibility of applying the results from an internally valid experiment to other situations and other research participants.
We logically design our research to rule out as many alternative interpretations of our findings as possible and to have any new facts be applicable to as wide a variety of other situations as possible. In many real-life situations in which external validity is high, however, it is impossible to rule out alternative interpretations of our findings. In a similar way, in laboratory settings in which internal validity is high, the setting is often artificial, and in many cases our findings cannot be generalized beyond the laboratory. Consequently, designing and conducting research is always a trade-off between internal and external validity. Which one we emphasize depends on the particular research questions being asked.
Before continuing, let’s clear up one misconception. It is the idea of designing “the one perfect study.” Although we strive to design good research, there are always alternative explanations and conditions not included in any single study. It is for this reason that Donald Campbell, who introduced scientists to the idea of internal and external validity, also emphasized the importance of replicating studies. If the same study is performed a number of times with similar results, then we can have more assurance that the results were valid. Even better, if the study is performed in a variety of settings around the world, we have even more confidence in our results. I will return to this topic of replication later in the chapter.
What Do I Expect to Happen?
One characteristic of human beings is that we seek to determine what will happen next. When we are talking with someone, we anticipate the next word they will say. The same is true in psychological experiments. Participants imagine what they are expected to do. If their expectation interferes with the influence of the IV, then the study could give inaccurate results.
In research terminology, this type of bias is referred to as demand characteristics. Demand characteristics occur when a participant’s response is influenced more by the research setting than by the IV. For example, a study might examine the effects of a drug as compared with exercise on reducing hyperactivity in adults. Many participants might believe the drug to be most effective, especially if the drug was given by a mental health professional and the exercise by a non–mental health professional. Those who received the drug might look for signs the drug was reducing hyperactivity. They might then pay more attention to the task used as the DV. If demand characteristics play an important role in the experiment, then they pose a significant threat to internal validity and offer an alternative explanation for understanding the influence of the IV.
demand characteristics: bias that occurs when a participant’s response is influenced more by the research setting than by the independent variable (IV)
placebo effect: the phenomenon that some people show psychological and physiological changes just from the suggestion that a change will take place
experimenter effects: bias that occurs due to the experimenter’s expectations
double-blind experiment: research procedure in which participants do not know whether they are in the experimental group or the control (placebo) group, and the researchers involved in the study also do not know which participants are in which group
blind controls: research participants who do not know whether they are in the experimental group or the control (placebo) group
A related phenomenon is referred to as the placebo effect. The term placebo comes from the Latin verb placere, which means “to please.” It refers to the phenomenon that some people show psychological and physiological changes just from the suggestion that a change will take place. How might this occur in a treatment study to reduce anxiety?
Experimenters also have expectations. For example, knowing that one set of subjects has been assigned to one condition rather than another could result in those participants being treated differently. Such situations are referred to as experimenter effects.
To control for the placebo effect in research as well as experimenter effects, various procedures have been used. One is to use a control group that receives either no treatment or a treatment previously shown to be ineffective for the particular disorder under study. In medical research, it is common to give a “sugar pill” that looks exactly like the medicine with the active ingredient. In psychotherapy research, a control group could be given relaxation training, which has been shown not to be effective on its own for mental disorders.
A more powerful control is to use a double-blind experiment in which the experimental group is divided into two groups. One group is given the actual treatment, and the other is given a treatment exactly like the experimental treatment but without the active ingredient. Neither the placebo group nor the experimental group would know which treatment they are receiving, and in this way, these research participants are said to be blind controls. The term double blind indicates that the experimenters giving the medication also do not know which treatment is experimental and which is a placebo.
In order to sort through the results of our experiments, we must be like detectives who constantly ask if there is another way to understand what was found. Our way of doing that is through research, logic, and doubt. We use research to design a study to consider alternative possibilities. We use logic to consider if our conclusions follow from the results. We use doubt to ask if there is a way to know if we are wrong.
Concept Check
What is the difference between the experimental group and the control group? Why is the control group important?
What is an operational definition? How does it help advance scientific knowledge?
How are the IV, the DV, and confounding variables related?
What is a double-blind experiment, and what are we trying to control for by using that design?
Designing an Experimental Study
One goal of experimental research is to determine the relationship between the IV and the DV. Less bias in terms of demand characteristics related to both the participant and the experimenter aids in creating a logical relationship between the IV and DV.
There are an additional four factors that are critical to sound inference (Figure 3.7). These are as follows:
1 Participant selection
2 Participant assignment
3 Design of experiment
4 Interpretation of relationship of IV to DV
Participants in the Study
It may seem simple to select subjects for an experiment. However, this is often a greater problem than it appears. The individuals selected for the study should be directly related to the hypothesis being tested. If you want to know if individuals with schizophrenia perform better on a memory test after drinking coffee, then of course you select individuals with schizophrenia. If you want to be
