The effects of instructions on subjects' disclosure of information about operant tasks.
Another important influence on the research methods used by behavioral scientists is the methodology of the physical sciences. Despite the differences in subject matter among physical and behavioral sciences, much of the methodology used by physical scientists is adopted by behavioral scientists without modification (Orne, 1962). Because behavioral scientists deal with animate organisms rather than inanimate objects, Orne emphasized the importance of studying what the subject does in the experiment rather than what is done to the subject. Consequently, the use of a certain procedure by an experimenter does not guarantee its effects on a subject's behavior.
One example of such a procedure involves the specific instructional practices used to minimize the effects of extraexperimental variables on the subject's behavior (Morris, Johnson, Todd, & Higgins, 1988). Human subjects enter an experiment with complex histories to which researchers have only indirect access (Morris et al., 1988; Weiner, 1969). The behavioral history of each person is subject to continual change between experimental sessions because of interactions with the extraexperimental environment (Pilgrim & Johnston, 1988). Ideally, experimental contingencies will be powerful enough to overcome other influences. However, the possibility remains that changes in behavior during an experiment might be the result of variables other than experimental manipulations (Morris et al., 1988). In an effort to reduce the effects of these uncontrolled stimuli, verbal instructions are commonly used (Baron & Galizio, 1983).
Instructions seem to have strong effects on behavior (Galizio, 1979). Even instructions that do not lead to efficient responding may exert influence if they do not lead to loss of reinforcement. The wording of instructions and the delivery of the instructions can also affect responding (Buskist, Bennett, & Miller, 1981; Catania, Matthews, & Shimoff, 1982). It is important, then, that researchers have evidence of the effectiveness of these instructions rather than simply assuming their effectiveness.
Another important consideration that has received inadequate empirical consideration in human operant research is the subject pool (Morris et al., 1988). In particular, because human operant research often employs undergraduate volunteers, subjects are likely to have contact with each other outside of the experimental situation. Thus, information about the experimental task could be transmitted to potential subjects by currently participating subjects. As Rosenthal (1966) points out, this transmission of information between subjects can influence subjects' behavior in the experimental setting. Despite the threat this poses to the validity of human operant research, the issue has been largely overlooked in the operant literature.
Because the experimental situation is a social interaction between experimenter and subject and because any transmission of information about the experiment to potential subjects can have serious consequences, it is necessary to understand the effectiveness of instructional practices regarding information disclosure in social contexts. In an attempt to examine the effectiveness of specific instructional practices, the present study addressed the following questions: (a) Does the specificity of the instructions affect subjects' disclosure of information? Specifically, does a statement of the possible negative effects of information disclosure increase the effectiveness of the instructions? (b) How does witnessing someone else disclose or refuse to disclose information affect subjects' disclosure? and (c) Does information disclosed to one subject by another actually affect the other subject's responding in the experimental session?
This experiment examined the effect of specificity of instructions on the disclosure of information concerning an experimental task. Some instructions contained a statement of the negative effects that disclosing information could have on the outcome of the experiment; other instructions merely requested that the subject not disclose, but made no reference to any specific effects of disclosure. This study also examined the effects of a more complex social situation on the effectiveness of instructions. Some subjects witnessed a confederate disclose or refuse to disclose information about an experiment before being asked to disclose information about her own participation.
Thirty Auburn University undergraduate female volunteers were randomly assigned to one of five groups. Subjects were recruited from introductory psychology classes to ensure minimal experience in behavioral research. All subjects received extra course credit for their participation.
Each subject was seated in a sound-attenuating cubicle (3 m square by 5 m high) facing an IBM personal computer. At the conclusion of the task, the subject was seated next to a female confederate in a closed waiting room containing three chairs and a table.
Upon arriving at the laboratory, each subject was given an informed consent form to read and was seated in the waiting room so that she would be familiar with this room during the final part of the experiment. After the subject signed the form, the experimenter read instructions to her regarding the disclosure of information about the experiment. A typed copy of the instructions remained with the subject during the experimental task.
Subjects in the detailed-instructions (DI) group and the witness-disclosure (WD) group were read the following:
Thank you for choosing to participate in this study. In a moment you will be seated in front of a computer. Instructions will appear on the screen that will tell you what to do. Please read them carefully and notify the experimenter when the computer tells you the task is over. Because some of your student colleagues may also volunteer to be subjects, we ask that you not discuss details of the experiment with them. Disclosing information about the experiment to other subjects could affect the experiment.
Subjects in the general-instructions (GI) group were read the following instructions:
Thank you for choosing to participate in this study. In a moment you will be seated in front of a computer. Instructions will appear on the screen that will tell you what to do. Please read them carefully and notify the experimenter when the computer tells you the task is over. We ask that you not discuss details of the experiment with anyone.
Subjects in the no-instructions (NO-I) group and the witness-refusal (WR) group were told:
Thank you for choosing to participate in this study In a moment you will be seated in front of a computer. Instructions will appear on the screen that will tell you what to do. Please read them carefully and notify the experimenter when the computer tells you the task is over.
Each subject was then led to the sound-attenuating cubicle and seated in front of the computer where she performed a matching-to-sample task. A geometric configuration appeared on the screen for one second and then disappeared. After a 2-3 s delay, the second configuration appeared for 1 s. The subject pressed the "K" key if the two configurations matched or the "1" key if they did not. After each response the computer displayed either "correct" or "incorrect" to indicate the subject's accuracy. The subject performed this task for 20 minutes each day for 3 consecutive days.
At the conclusion of the last session, each DI, GI, and NO-I subject was taken to the waiting area where a female confederate was already seated. The experimenter introduced the confederate as a subject waiting to begin the study and gave her a consent form to read and sign. The experimenter then excused himself to complete the subject's extra-credit form and left the waiting area, closing the door behind him.
Once the experimenter left the room, the confederate asked the subject to describe the experiment she had just completed. The confederate asked the subject "Did you just finish the experiment? What did you have to do?" The conversation between the two was audiotaped to ensure an accurate record of the information disclosed by the subject. When the conversation about the experiment had ended, the confederate signed the consent form and then left to find the experimenter. This prevented the experimenter from returning to the waiting area prematurely and interrupting the interaction.
For the WD and WR groups, each subject was escorted to the waiting area where two confederates were already seated. The experimenter introduced one confederate as a current subject in an experiment (Confederate A) and the other as a subject who was about to begin the experiment (Confederate B). The experimenter then gave Confederate B a consent form and excused himself to complete the real subject's extra-credit form.
Subjects in the WD group, who received the detailed instructions, witnessed Confederate B ask Confederate A to divulge details of her experiment. Confederate A told Confederate B, "1 had to press the space bar on a keyboard to earn points on a counter." Confederate B then asked the real subject to reveal the details of her experiment.
Subjects in the WR group, who received no instructions regarding information disclosure, also witnessed Confederate B ask Confederate A about her experiment. However, in this case, Confederate A told Confederate B that she "was not allowed to discuss the experiment with anyone because it might affect the experiment." Confederate B then asked the subject to tell about her experiment. When the conversation ended, the confederate signed the form and then left the room to find the experimenter. These conversations also were audiotaped.
Results and Discussion
A written transcript of each conversation was prepared from the audiotapes and the amount of disclosure was rated as complete, partial, or none by eight undergraduate volunteers (six female, two male) who were blind to the purpose of the experiment. Complete disclosure was defined as the discussion of the stimuli on the screen and the matching procedure. Partial disclosure was the mention of some, but not all, of the components of the experimental task. For example, telling the confederate that shapes appeared on the screen was classified as partial disclosure because no mention of matching the sample stimuli to comparison stimuli was made. A conversation was classified as containing no disclosure if the subject said nothing of the shapes or the matching procedure. Because the informed consent forms stated that a computer task was involved, interactions that only mentioned a computer were not considered to be disclosing.
The degree of disclosure for individual subjects in each of the groups is shown in Table 1. Statistical analysis of the DI, GI, and NO-I subjects' responses showed a significant relationship between the type of instruction and the degree of disclosure ([[Chi].sup.2] = 14.99, p [less than] .01).
In the DI group, five of the six subjects did not disclose any significant information about the experimental task. One subject stated that two keys were used for different things, resulting in a rating of partial disclosure. In both the GI and NO-I groups, five of six subjects made full disclosures of the matching procedure. One subject in each group made a partial disclosure. The GI subject described shapes on the screen, but did not mention the matching. The NO-I subject stated that she matched things, but said nothing of the stimuli used or of the specific matching procedure.
Number of Subjects Making Each Type of Disclosure in Experiment 1
Type of Disclosure Group Complete Partial None
DI 0 1 5 GI 5 1 0 NO-I 5 1 0 WD 3 3 0 WR 1 3 2
The detailed instructions were more effective in preventing disclosure. Interestingly, subjects receiving the general instructions responded in a fashion similar to subjects who had received no instructions, suggesting that the statement of consequences contained in the detailed instructions exerted control over the subjects' disclosure of information. Without that statement, the instructions appeared to be ineffective.
Three subjects in the WD group made a complete disclosure; three made a partial disclosure. Although the DI and WD groups received the same instructions (the only difference being the witnessing of a confederate disclose information in the WD group), the verbal behavior of these groups were analyzed separately. Despite receiving the instructions, which seemed to prevent almost all information disclosure in the DI group, three subjects in the WD group made complete disclosures after witnessing a confederate disclose information ([[Chi].sup.2] = 9.0, p [less than] .02).
In the WR group, one subject made a complete disclosure, three made partial disclosures, and two disclosed no information. Although the NO-I and WR groups received identical instructions, the behavior of subjects in each group was different. In the NO-I group, five subjects made complete disclosures and one made a partial disclosure. However, when subjects in the WR group witnessed a confederate refuse to disclose information, only one complete disclosure was made ([[Chi].sup.2] = 5.66, p [less than] .10). Two subjects in the WR group actually refused to disclose any information about the task, even though they had never been instructed to do so.
Experiment 1 demonstrated that, without special precautions, subjects tend to divulge information about their participation in an experiment. Thus, it is possible that subsequent subjects drawn from that same subject pool might perform partly under undesired instructional control. To examine this possibility, Experiment 2 addressed the effects of information disclosure on subjects' responding on an operant task.
Eight Auburn University undergraduate female volunteers were randomly assigned to one of two groups. Subject recruitment was the same as described in Experiment 1.
The experimental apparatus used was the same as in Experiment 1.
Upon arriving at the laboratory, each subject was taken to the waiting room by the experimenter and seated next to a confederate. After the experimenter left the room, the confederate introduced herself as a subject who had just finished the experiment. She then told four subjects that she had "pressed the space bar once every 20 seconds." She told the other four subjects that she "pressed the space bar as fast as possible."
Once this information was disclosed, the confederate excused herself and left the room. The experimenter then returned to the waiting room and escorted the subject to the experimental chamber where she was seated in front of the computer. The following instructions appeared on the monitor, "Press the space bar to earn points." Points were delivered on a fixed-interval (FI) 20-second schedule for one 20-minute session. Upon completion of the schedule requirements, 25 points were added to the subject's total. A cumulative point total was displayed on a counter on the monitor.
Results and Discussion
Each subject's postreinforcement pauses (PRP) were recorded for each interreinforcement interval. These values were averaged for blocks of 10 intervals resulting in four blocks for each subject as shown in Figure 1. Response rates were also calculated and are shown in Figure 2.
The disclosure of information by a confederate to subjects clearly affected responding. Each subject's responding was consistent with the confederate's fictitious description of her responding. Subjects given the "respond every 20 seconds" report displayed PRPs of approximately 20 seconds in the initial intervals and maintained that tendency throughout the session. The response rates for these subjects were relatively low, averaging 5.2 responses per minute across subjects.
In contrast, subjects to whom the confederate reported responding as quickly as possible displayed relatively short PRPs and high response rates, averaging 131.8 responses per minute.
It is important to note that, although the results of this study are consistent with other studies of instructional effects (e.g., Buskist et al., 1981), the subjects in this experiment were not specifically instructed to respond in a certain way; they only heard the confederate disclose information about her responding. However, the disclosure of information seems to affect behavior much like instructions. The observed effects on the subjects' FI responding suggest that this type of disclosure is a real threat to experimental control.
The results of these studies may be summarized as follows: (a) Instructions describing negative consequences of disclosing information were most effective in preventing disclosure, (b) instructions requesting subjects not to disclose information without stating consequences were relatively ineffective, (c) instructional effects were influenced by the social interactions witnessed by the subjects, and (d) information disclosed to subjects affected responding. As long as the process of subject selection allows contact between past or current subjects and future ones, researchers need to minimize the possibility of information being leaked to potential subjects.
Because instances of disclosure are social in nature, some relevant issues have been addressed in traditional social psychological research. One topic addressed in this research, related to the effects of witnessing others disclose or refuse to disclose information, is the influence of norms, certain accepted rules of social behavior (Myers, 1993). Although norms develop in common or familiar circumstances, they may be weak or nonexistent in novel or unfamiliar situations. In contexts for which no norms exist, people commonly imitate, or model, the observed behavior of those around them. For example, Sherif's (1937) studies of the autokinetic effect examined the emergence of group norms in an ambiguous situation. When subjects were asked to estimate, aloud, the distance a stationary light appeared to move, initial responses were diverse. However, over time the subjects' responses began to converge and a group norm was established. Each subject's response was increasingly influenced by the responses of the other subjects leading to a definite, although false, consensus.
The potency of social factors in influencing individual behavior is especially evident in Asch's (1956) classic study of conformity. A subject and six confederates were shown a standard line and three comparison lines. The experimenter asked them to say which comparison line was closest in length to the standard line. The three comparison lines were clearly different in length, making the correct response obvious. However, each confederate had been instructed to give a particular incorrect response. Because the subject was seated in the sixth position, giving the correct response required nonconformity (going against a unanimous majority). Although 99% of the responses were correct when a subject responded alone, only 63% of the responses were correct when nonconformity was required. Whereas Sherif used an ambiguous situation, Asch discovered that social pressures to conform were powerful enough to make people respond in direct contradiction to the actual contingencies.
Because the experimental setting is not a common environment for most undergraduate volunteers, few adequate social norms exist. This places the subject in a situation similar to those in the Sherif and Asch studies. Consequently, subjects' behavior is likely to be influenced by the behavior of those around them. The powerful control of social variables described by Asch and Sherif should not be ignored in this context.
There are other variables that may be relevant to information disclosure but were not addressed by this study. One such factor is the effect of any preexisting relationship between two people on the probability of one disclosing information to the other. The confederates used in this study were strangers to the subjects. Subjects may respond differently if asked to disclose information by someone they know well, such as a friend, roommate, boyfriend, or girlfriend. The laboratory setting used in this study may also have influenced the subjects' responses. The greater security and privacy of a subject's own dwelling may increase the likelihood of information disclosure.
Baron and Galizio (1983) have argued that human behavior cannot be assumed to be under sole control of contingencies whenever instructions are used. Given the similar effects of disclosed information on human responding, the same argument can be applied to any experimental situation in which disclosure may have occurred. As long as there is the possibility that information regarding an experiment is disclosed to a subject, that subject's behavior cannot be assumed to be under sole control of experimental contingencies. This is particularly important in short-term experiments where the subject has less contact with the experimental contingencies.
The specific laboratory practices used by researchers are attempts to ensure experimental control and gain an empirical and scientific understanding of human behavior. However, because the use of these practices constitute behavior, these practices must be examined as rigorously as any other behavior. What has long been absent in the experimental analysis of human behavior is programmatic methodological research. This emphasis is evident in a census of human operant behavior literature from 1982 to 1992, which indicates that human operant researchers have focused primarily on content-oriented issues (Dougherty, Nedelmann, & Alfred, 1993).
The present work shows that human operant researchers should not ignore methodologically oriented research and suggests at least one variable, subject disclosure of information, that may compromise the integrity of studies using human subjects. Researchers using human subjects need to present subjects with specific instructions regarding the potentially damaging effects of disclosing information. Similarly, researchers should attempt to minimize the opportunities for potential subjects to ask previous subjects about the experiment. This could be accomplished, at least in part, by minimizing the number of subjects recruited from the same class.
Although methods such as random assignment and stability requirements can minimize the effects of some extraneous variables, one cannot assume that they are sufficient safeguards against the effects of the disclosure of information. Clearly, more attention and research needs to be directed toward understanding the effectiveness of the procedures used with human subjects.
ASCH, S. E. (1956). Studies of independence and conformity: A minority of one against a unanimous majority. Psychological Monographs, 70, (9, Whole No. 416).
BARON, A., & GALIZlO, M. (1983). Instructional control of human operant behavior. The Psychological Record, 33, 495-520.
BUSKIST, W. F., BENNETT, R. H., & MILLER, H. L. (1981). Effects of instructional constraints on human fixed-interval performance. Journal of the Experimental Analysis of Behavior, 35, 217-225.
CATANIA, A. C., MATTHEWS, B. A., & SHIMOFF, E. (1982). Instructed versus shaped human verbal behavior: Interactions with nonverbal responding. Journal of the Experimental Analysis of Behavior, 38, 233-248.
DOUGHERTY, D. M., NEDELMANN, M., & ALFRED, M. (1993). An analysis and topical bibliography of the last ten years of human operant behavior: From minority to near majority (1982-1992). The Psychological Record, 43, 501-530.
GALIZIO, M. (1979). Contingency-shaped and rule-governed behavior: Instructional control of human loss avoidance. Journal of the Experimental Analysis of Behavior, 31, 53-70.
GALIZlO, M., & BUSKIST, W. F. (1988). Laboratory lore and research practices in the experimental analysis of human behavior: Selecting reinforcers and arranging contingencies. The Behavior Analyst, 11, 65-69.
MORRIS, E. K., JOHNSON, L. M., TODD, J. T., & HIGGINS, S. T. (1988). Laboratory lore and research practices in the experimental analysis of human behavior: Subject selection. The Behavior Analyst, 11, 43-50.
MYERS, D. G. (1993). Social psychology (4th ed.). New York: McGraw-Hill.
ORNE, M. T. (1962). On the social psychology of the psychology experiment: With particular reference to demand characteristics and their implications. American Psychologist, 17, 776-783.
PILGRIM, C., & JOHNSTON, J. M. (1988). Laboratory lore and research practices in the experimental analysis of human behavior: Issues in instructing subjects. The Behavior Analyst, 11, 59-64.
ROSENTHAL, R. (1966). Experimenter effects in behavioral research. New York: Appleton-Century-Crofts.
SHERIF, M. (1937). An experimental approach to the study of attitudes. Sociometry, 1, 90-98.
WEINER, N. (1969). Conditioning history and human avoidance and escape responding. Journal of the Experimental Analysis of Behavior, 12, 1039-1043.
|Printer friendly Cite/link Email Feedback|
|Author:||England, Dan E.; Buskist, William|
|Publication:||The Psychological Record|
|Date:||Jun 22, 1995|
|Previous Article:||Stimulus equivalence classes involving olfactory stimuli.|
|Next Article:||The effects of schedules of reinforcement on instruction following.|