One advantage that physicists and chemists have over psychologists when it comes to conducting research is that the “subjects” of physical and chemical experiments (e.g., balls rolling down inclined planes) pay absolutely no attention to the fact that they are participants in an experiment. Th ey behave as they ordinarily do in nature. Th e sub- jects (rats, pigeons) and participants (college students, human adults) of psychological research do pay attention to their status as such and may modify their behavior as a result of their perceptions. Th is “reactive” nature of subjects and participants must be considered when designing and assessing psychological research. Th is section describes the kinds of reactions, the situations that sometimes give rise to them, and the things you can do to minimize (or at least assess) their impact on your data. A discussion of research with human participants begins this section, followed by a discussion of research with animal subjects.
Reactivity in Research with Human Participants
Assume for the moment that you have defi ned your population of participants and are now ready to acquire your participants and run your experiment. You plan to have volunteers sign up and come to your laboratory for your experiment. What can you expect from these creatures that we call human participants?
One thing to realize is that the psychological experiment is a social situation. You as the experimenter, by defi nition, are in a position of power over the participant. Your participant enters this situation with a social history that may aff ect how he or she responds to your manipulations. Assuming that the participant is a passive recipient of your experimental manipulations is a mistake. Th e participant is a living, think- ing human being who will generate personal interpretations of your experiment and perhaps guide behavior based on these interpretations. In short, the participant is a reactive creature. Th e behavior that you observe in your participants may not be repre- sentative of normal behavior simply because you are making observations.
To help you understand the reactions of research participants to your experiment, imagine that you have volunteered for a psychological experiment for the fi rst time.
You are a fi rst-year student enrolled in introductory psychology who has had a little experience with psychological research. As you sit waiting to be called for the experi- ment, you imagine what the experiment will be like. Perhaps you have just talked about Milgram’s obedience research in your psychology class or saw a documentary about it on television and are wondering if you are going to be given electric shocks or if the researcher is going to be honest with you. You wonder if you are going to be told the experiment is about one thing when it is actually about something else.
At last, the experimenter comes out of the laboratory and says she is ready for you. You are led into a room with a white tile fl oor, white walls, a stainless steel sink in the corner, some ominous-looking equipment in another corner, and rather harsh fl uorescent lighting. Th e experimenter apologizes for the cold surroundings and says that she is a graduate student and had to settle for one of the animal labs to run her master’s thesis research. (Do you believe her?) You take a look around the room and
146 CHAPTER 5 . Making Systematic Observations
muster enough courage to ask whether you are going to be shocked. Th e experimenter chuckles and assures you that the experiment deals with memory for abstract pictures.
She then begins to read you the instructions. At that moment, some workers begin to hammer out in the hall. (Is this part of the experiment?) Again, the experimenter apologizes. She explains that they are installing a new air-conditioning system in that wing of the building. You think you detect a hint of a smile on her face. You don’t believe her. You have decided that the experimenter is really trying to test how well you can perform a memory task under distracting conditions. You decide to “show the experimenter” that you can do well despite her obvious attempt to trick you. Th e experimenter runs you through the experiment. Of course, you try your hardest to get all the items right.
After you have completed the memory test, the experimenter asks you whether you have any questions. You smugly tell her that you saw through her obvious decep- tion and worked even harder to get the items correct. After all, you weren’t born yester- day! To this the experimenter incredulously assures you that the noise was not part of the experiment and tells you that she will have to throw out your data. Th e experiment has been set back an entire day.
Demand Characteristics
Consider the psychological experiment in the light of this example. As stated, the human participant in a psychological experiment does not passively respond to what you may expose him or her to. On entering the laboratory, your participant probably assesses you and the laboratory (Adair, 1973). Given these assessments, the participant begins to draw inferences concerning what the experiment is about.
Th e cues provided by the researcher and the context that communicate to the participant the purpose of the study (or the expected responses of the participant) are referred to as demand characteristics. Participants gain information about the experi- ment from these demand characteristics. Unfortunately for you, the participant may be paying attention to cues that are irrelevant to the experiment at hand (as hap- pened in the previous example when you believed the noise created by the work crew was related to the experiment). With the information obtained from the demand characteristics, the participant begins to formulate hypotheses about the nature of the experiment (such as “Th e experiment is measuring my ability to perform under adverse conditions”) and begins to behave in a manner consistent with those hypoth- eses. Problems occur when the participant’s hypotheses diff er from the intended pur- pose of the experiment. Adair (1973) refers to this class of demand characteristic as
“performance cues.”
A second source of demand characteristics centers on the participant. According to Adair (1973), a class of demand characteristics known as role attitude cues may sig- nal the participant that a change in the participant’s attitude is needed to conform to his or her new role as a research participant (Adair, 1973, p. 24). Further, Adair points out that participants enter experiments with preexisting attitudes that dispose them to react in either a positive or a negative way to the experimental manipulations. Th rough various demand characteristics, the experiment can cause the participant to change his or her attitudes (Adair, 1973, p. 26). Adair lists the following three categories of
bor35457_ch05_123-157.indd 146
bor35457_ch05_123-157.indd 146 4/22/13 2:52 PM4/22/13 2:52 PM
THE REACTIVE NATURE OF PSYCHOLOGICAL MEASUREMENT 147 predisposing attitudes of participants: the cooperative attitude, the defensive or appre-
hensive attitude, and the negative attitude.
Th e cooperative attitude is characterized by a strong desire to please the experi- menter. According to Adair, volunteering for an experiment “seals a contract between the experimenter and the participant, fostering cooperative behavior” (1973, p. 26).
Reasons for the cooperative attitude include a desire to help science, a desire to please the experimenter, a desire to perform as well as possible, and a desire to be positively evaluated by others.
Several demonstrations of the impact of this positive attitude on the outcome of an experiment have been made. For example, Orne (1962) demonstrated that par- ticipants will engage in a boring, repetitive task for hours to please the experimenter.
Participants were provided 2,000 sheets of paper (on which were columns of numbers to add) and a stack of index cards (on which instructions were printed). Th ey were instructed to select the fi rst card (which told the participants to add the numbers on the page) and then to select the next card. Th e next card told the participants to tear the completed sheet into pieces (not fewer than 32) and then select another sheet and add the numbers. Th e cycle of adding numbers and tearing sheets continued for as long as a participant was willing to go on.
If you were a participant in this experiment, what would you do? You may have said, “I’d do it for a few times and then quit.” In fact, quite the opposite happened:
Participants continued to do the task for hours. Evidently, participants perceived the test as one of endurance. Th e participants’ cooperative attitude in this example inter- acted with the demand characteristics to produce some rather bizarre behavior. Th is
“good participant” eff ect was also shown in an experiment by Goldstein, Rosnow, Goodstadt, and Suls (1972).
Some participants enter the laboratory worried about what will happen to them and have an apprehensive attitude. One of us (Bordens) was conducting an experi- ment on jury decision making, and several participants, on entering the lab, asked if they were going to be shocked. Th is apprehension may stem from the partici- pants’ perception of the experimenter as someone who will be evaluating the par- ticipants’ behavior (Adair, 1973). Th e apprehensive attitude also has been shown to aff ect the research outcome, especially in the areas of compliance and attitude change (Adair, 1973).
Some participants come to the laboratory with a negative attitude. Even though most participants are either positive or defensive (Adair, 1973), some participants come to the lab to try to ruin the experiment. Th is attitude was most prevalent when participants were required to serve in experiments. Required participation made many participants angry. Th e present rules against forced participation may reduce the fre- quency of negative attitudes. However, you cannot rule out the possibility that some participants will be highly negative toward the experiment and experimenter.
Other Infl uences
In addition to demand characteristics and participants’ attitudes, evidence also indi- cates that events outside the laboratory can aff ect research. For example, Greene and Loftus (1984) conducted an experiment on jury decision making in which eyewitness
148 CHAPTER 5 . Making Systematic Observations
testimony was being studied. Around the time that the experiment was conducted and in the same city, a celebrated case of mistaken identifi cation was being unmasked.
Knowledge of that case was refl ected in the data. Participants generally were more skeptical of the eyewitness in the study after fi nding out about the case than they were before. However, after a while, the impact of the case diminished, and the responses returned to “normal.”
Th e moral to this story is that participants are not passive responders to the exper- iment. Th e experiment is a social situation in which the interaction between partici- pant attitudes and the experimental context may aff ect the outcome of the experiment.
As a researcher, you must be aware of demand characteristics and take steps to avoid them or at least to assess their impact. As with other participant-related problems, demand characteristics, participant attitudes, previous research experience, and expo- sure to everyday life can aff ect both internal and external validity.
Th e Role of the Experimenter
Th e participant is not the only potential source of bias in the psychological experi- ment. Th e experimenter can sometimes unintentionally aff ect the outcome of the experiment. Assume that you are running your fi rst experiment, an experiment of your own design. Because you are a student, you will be testing your own participants.
You are sitting alone in your laboratory, awaiting the arrival of your fi rst par- ticipant. You have butterfl ies in your stomach and are a bit apprehensive about how you will perform in the experiment. Th e experiment is important to you because it is required for a class that you need for graduation. At last, your fi rst participant arrives, and you usher him into your laboratory. You begin to read your instructions (which you feel are well written) to your participant and are puzzled to see that your partici- pant is obviously not understanding the instructions. However, you press on.
Your experiment deals with the ability of people to recall certain words embed- ded within the context of other words. You want to show that interference will occur when the words are embedded in a context of other similar words. You are going to read a list of words to your participant and then give a recall test. In the high-similarity condition, you unconsciously read the words at a faster rate than in the low-similarity condition. You notice later that your collected data consistently confi rm your preex- perimental hypothesis.
Now, analyze what has happened. You wrote your instructions, believing that your participants would be able to understand them. As it turns out, the instruc- tions were less clear than you thought. Th e problem here was that you assumed too much about the ability of your participants to understand the instructions. Th is may happen because you are used to talking to other psychology majors or profes- sors familiar with the jargon of your discipline. Th e participants may not have that advantage. One thing that you could do to detect this problem is to pretest the instructions.
Experimenter Bias In the classic 1960s sitcom Mr. Ed, Wilbur Post owned a horse named Ed with a special talent: Ed could talk. In each episode, Ed’s antics created some interesting problems for his owner Wilbur. Over a half century before Mr. Ed
bor35457_ch05_123-157.indd 148
bor35457_ch05_123-157.indd 148 4/22/13 2:52 PM4/22/13 2:52 PM
THE REACTIVE NATURE OF PSYCHOLOGICAL MEASUREMENT 149 hit the airwaves, another horse, named Hans, created a sensation in the entertainment
world in Europe. Hans, it seemed, could solve simple mathematical problems. His owner, Wilhelm Von Osten, took great pains to teach Hans to solve the problems and then took Hans on the road to entertain people. Von Osten would show Hans a card with a math problem (e.g., an addition problem), and Hans would begin clopping out the answer with his hoof. Hans would stop clopping his hoof when the correct answer was reached. Audiences were astounded and for two years Hans earned Von Osten a nice living.
Not everyone was taken with Hans’s mathematical prowess. A scientist named Oskar Pfungst doubted that Hans was able to solve math problems. Instead, he believed that Hans was picking up subtle cues from Von Osten. So Pfungst designed a series of tests to see if Hans had the miraculous abilities claimed. In one test, Von Osten showed Hans a card with a problem. Th e catch was that Von Osten did not know what the problem was. In this and similar tests in which Von Osten was not allowed to see the problem being put to the horse, Pfungst found that Hans could not solve the problems. Pfungst believed that Hans was reading his trainer’s behavior, looking for cues to signal when Hans should stop clopping his hoof. In fact, Pfungst found that as Hans began clopping, Von Osten would unconsciously tense up, which showed in his body position and facial expressions. As Hans reached the correct answer, Von Osten would unconsciously relax, signaling Hans that the correct answer had been reached (Wozniak, 1999). In cases where Von Osten did not know what problem Hans was to solve, Von Osten could not provide Hans with the unconscious signals, and Hans’s performance deteriorated.
At this point, you may be asking yourself, “What does this have to do with my research? I don’t plan on dragging a horse around to entertain people.” Whoa! Let’s slow down and see how the case of Hans relates to your research. In fact, it relates in a quite simple way. Th e “Clever Hans phenomenon” poses a potential threat to the valid- ity of your research. Let’s take a look at a modern-day research example to see how this might work.
Th ere is a phenomenon known as facilitated communication, which involves a “facilitator” physically helping an impaired person communicate by touching let- ters on a screen. Th e facilitator supports the impaired person’s hand while the person guides his fi nger to a symbol on a screen (Montee, Miltenberger, & Wittrock, 1995).
Supposedly, this technique allows the impaired person to communicate with others in ways and at levels previously believed to be impossible. But is facilitated communica- tion a real phenomenon or another example of the Clever Hans phenomenon? Let’s fi nd out.
Barbara Montee, Raymond Miltenberger, and David Wittrock (1995) con- ducted an experiment to test the validity of facilitated communication. Seven cli- ent–facilitator dyads participated in this experiment. Th e experiment was conducted in the client–facilitator pairs’ normal setting (e.g., day-care center) at the usual time of day. Th e pairs completed several facilitated communication tasks involving describ- ing an activity or naming a picture. Th e independent variable was the information provided to the facilitators prior to the facilitated communication session. In one condition (the known condition), the facilitator was informed about the activity that the client had engaged in or the picture that the client had been shown. In another
150 CHAPTER 5 . Making Systematic Observations
condition (the unknown condition), the facilitator was not informed about the activ- ity or picture. In the fi nal condition (the false feedback condition), the facilitator was given incorrect information about the activity or picture. Th e dependent variable was whether, using facilitated communication, the client correctly described the activity or named the picture.
Th e results from this experiment were rather dramatic and are shown in Table 5-1 . As you can see, the client’s ability to describe the activity or name the picture was almost totally dependent upon whether the facilitator had accurate information about the nature of the activity or picture. Just as Hans could not solve his math problems when Von Osten did not know the answer, so the clients in this experiment could not respond correctly unless the facilitators knew the answers.
In both the Clever Hans and facilitated communication situations, there was a common problem known as experimenter bias (Rosenthal, 1976). Experimenter bias creeps in when the behavior of the experimenter infl uences the results of the experiment. Th is infl uence serves to confound the eff ect of your independent vari- able, making it impossible to determine which of the two was responsible for any observed diff erences in performance on the dependent measure. Experimenter bias fl ows from at least two sources: expectancy eff ects and treating various experi- mental groups diff erently to produce results consistent with the preexperimental hypotheses.
When an experimenter develops preconceived ideas about the capacities of the participants, expectancy eff ects emerge. For example, if you believe that your par- ticipants are incapable of learning, you may treat them in such a way as to have that expectation fulfi lled. Rosenthal (1976) reports a perception experiment in which the independent variable was the information provided to students acting as experi- menters. Some students were told that, according to previous ratings, their partici- pants should perform well. Others were told that the participants would probably perform poorly. Th e student experimenters also were told they would be paid twice as much if the results confi rmed the prior expectations. Rosenthal reports that establishing the expectancy led to diff erent behavior on the part of the participants in the two experimental groups. Rosenthal points out that such expectancy eff ects may be a problem in not only experimental research but also survey research and clinical studies. In the previous hypothetical example, you (as the experimenter)
TABLE 5-1 Mean Number of Correct Responses Made in the Montee, Miltenberger, and Wittrock (1995) Experiment on Facilitated Communication
INFORMATION CONDITION
TASK Known Unknown False
Picture naming 75 0 1.8
Activity identifi cation 87 0 0
bor35457_ch05_123-157.indd 150
bor35457_ch05_123-157.indd 150 4/22/13 2:52 PM4/22/13 2:52 PM