Winter 1978, Volume 4, Number 1

The American Psychological Association (1973) report on ethical principles of research with human participants is the center of a current controversy. Several psychologists have commented on the restrictions researchers may experience with an explicit code of ethics. Steiner (1972) points out that a typical college examination may induce more stress than most psychological experimentation. Gergen (1973) feels that until scientific answers about the effects of deception, coercion, and informed consent are evident, strong APA policies seem unwarranted. However, Johnson (1974) feels that it is a basic contravention of human rights to set aside the possible effects of research until they are proved to be harmful.

Researchers have examined how samples from various populations rate psychological experimentation. Sullivan and Deiker (1973) compared subject and experimenter perceptions of representative ethical issues in human research. Four experiments with controversial design characteristics were used. In 18 of 20 response categories, psychologists gave a significantly stricter interpretation of the ethical issues than did the student subjects. There were two categories, both in conjunctions with the experiments in which subjects were asked to give or receive electric shock, where students gave more ethically stringent responses.

Wilson and Donnerstein (1976) assessed the general public's ratings of eight nonreactive field studies. Four of the eight studies were seen as harassing by a majority of the subjects, but in no case did a majority see the procedures as invading their privacy or unethical. The authors concluded that interpretations of these results are dependent upon the extent of one's concern with what the "majority" feels.

Berschied, Baron, Dermer, and Libman (1973) showed that a behavioroid measure of consent rates for participation in stressful experiments decreased as more procedural information was revealed. Consent rates were significantly lower when subjects were given procedural information or told that they were likely to act negatively.

What are the relevant characteristics of an experimental description which influence subjects' ratings? How does the general public weigh the costs and benefits of research? Do they place more emphasis on qualities of the procedure, or do they tend to emphasize what subjects did in response to the procedure when rating experimentation? It is the aim of the present study to provide some insight into these questions by examining how deception and outcome influence subjects' ratings of the still controversial Milgram (1964) obedience study.

Subjects and procedure. Subjects were 100 persons (50% male, 50% female) systematically sampled from in front of Chicago's Civic Center and from riders of the Rapid Transit lines. In the Civic Center, only people sitting on one of the four sides surrounding the Picasso sculpture were chosen. A coin toss decided the sex of the first four subjects to be sampled. Subjects five through eight were then of the opposite sex, with the coin toss occurring after every eighth subject. Each subject in a group of four was sampled from a different side of the sculpture. A marble chosen from the experimenter's pocket decided which of the four sides of the sculpture was to be sampled from first. A separate marble choice then decided from which of the three sections of a chosen side (left-end, middle, right-end) a subject was to be sampled.

On the subway, a coin toss again decided the sex of the first eight subjects to be sampled. After the sex of the subject had been decided, another coin toss decided whether the east or west side of the subway car was to be used for sampling. The first person of the appropriate sex seated alone, starting from the north end of the car, was asked to participate. Only one subject was run in each subway car.

In both the subway and the Civic Center, the order of presentation for the various conditions was also randomized, with each subject in a group of four being in a different condition. Twelve persons (eight males, four females) refused to participate in the study. The data from fourteen subjects (seven males, seven females) had to be discarded due to either forgetting to fill out the face sheet, or inappropriate use of the rating scales. Additional subjects were used to replace these subjects.

After a prospective subject agreed to participate, he or she filled out a bibliographical face sheet and read one of the four experimental descriptions used in this study.

Deception-high obedience. In this condition, subjects read that: A group of people volunteered to be in a research study at a prominent university. The researcher asked these volunteers to give "electric shocks" to a "learner" as part of a "learning experiment". The researcher also said that for each wrong answer made by the "learner" he would be ordering the volunteers to give stronger and stronger "shocks". Before the study began, the researcher did not tell the volunteers that he was really studying how obedient they would be when ordered to inflict pain on another human being.

Again, the volunteers did not know that the "learner" receiving the "shocks" was really an assistant of the researcher. The volunteers were also unaware that no real shocks would be given. Almost all the volunteers "shocked" the "learner" after he began to say that he was in pain. Most of the volunteers administered "shocks" labeled "very dangerous". After these results were obtained, the researcher informed the volunteers of the true purpose of the study.

Deception-low obedience. Subjects in this condition read the exact same abstract as given above, with the exception of a reversal in results. Subjects read that, "Almost all the volunteers refused to 'shock' the 'learner' after he began to say that he was in pain. Most of the volunteers did not administer 'shocks' labeled 'very dangerous'."

No deception-high obedience. In this condition, subjects read that: A group of people volunteered to be in a research study at a prominent university. The researcher asked these volunteers to make believe they were going to give "electric shocks" to his assistant as part of a "learning experiment". The researcher said that for each wrong answer made by his assistant, he will be ordering the volunteers to give stronger and stronger "shocks". Before the study began the researcher told the volunteers that he was really studying how obedient they would be when ordered to inflict pain on another human being.

Again, the volunteers knew that the "learner" receiving the "shocks" was really an assistant of the researcher. The volunteers were also aware that no real shocks would be given. Almost all the volunteers "shocked" the "learner" after he began to say that he was in pain. Most of the volunteers administered "shocks" labeled "very dangerous".

No deception-low obedience. Subjects in this condition read the same abstract as subjects in the no deception-high obedience condition, with the exception of a reversal in results. Subjects read that, "Almost all of the volunteers refused to 'shock' the 'learner' after he began to say that he was in pain. Most of the volunteers did not administer ‘shocks’ labeled "very dangerous’."

Results and discussion. After reading one of the above variants of the Milgram abstract, each subject responded to ten 8-point scales. The data were analyzed by means of multivariate analysis of variance. The results show no significant effects associated with deception (F(10,87) = 0.99, p < .46) nor any significant interactions of deception and level of obedience (F(10,87) = 0.78, p < .63). None of the biographical categories were significantly related to any of the subjects' ratings. However, the multivariate test of the effects of outcome on subjects' ratings showed a significant difference (F(10,87) = 4.53, p < .001).

A series of univariate analyses showed that subjects in the high obedience condition thought the results of the study were more unfavorable about human nature than subjects in the low obedience condition. Subjects in the high obedience condition rated the experiment as significantly more harmful to the volunteers than subjects in the low obedience conditions. High obedience condition subjects also felt the volunteers had a significantly worse experience and acted more irresponsibly than subjects reading the low obedience versions. However, subjects in the low obedience condition felt the volunteers were more uncomfortable in disobeying. The means of the various groups in relation to the above mentioned significant differences are shown in Table 1.

Table 1

The Means for the Combined "Outcome" Groups on Five Dimensions

*1 = unfavorable (F(1, 96) = 16.22, p < .001)
**1 = harmful (F(1, 96) = 3.85, p < .05)
***1 = good experience (F(1, 96) = 8.41, p < .001)
****1 = irresponsible (F(1, 96) = 23.31, p < .001)
*****1 = uncomfortable (F(1, 96) = 8.24, p <.05)

The univariate F's for the following questions were not significant. (1) If another researcher wanted to repeat this particular experiment, how right would it be for him to do so? (2) How worthwhile was this study? (3) How appropriate was it for this study to be financed with federal tax money the researcher received in a grant? (4) How valuable to science is this particular study? (5) How useful would you say the results of this study are? It should be noted that pilot testing indicated some ambiguity in subjects' interpretations of the question, "How ethical was this experiment?" Thus, this particular question was not used in this study. It should also be noted that 62% of all subjects in this study felt that other researchers should be encouraged to do more research on obedience.

The ratings of the subjects in this study were more contingent upon the outcome of the experiment rather than the particular procedure used to obtain the results. It is conceivable that the level of deception depicted in the abstracts did not exert a significant effect because the subjects did not notice that deception was being used by the researcher in the descriptions. However, this is unlikely since only 4% of all the subjects reading the deception variants thought that the volunteers were not deceived.

One may wonder if the Milgram study would have been the subject of public outrage if the results had turned out differently. These data suggest that if most of Milgram's subjects had disobeyed, his experiment would have not received as much condemnation.

It has been suggested that unless psychologists limit their own research with enforced ethical standards, others may do so more harshly. However, if the results of a study are a major criterion for judging the ethicality of a procedure, as this study suggests, then researchers must deal with the ethical objectives of their research from a different perspective. Critics may be responding to the unflattering portrayal of human nature discovered using deceptive methodologies rather than the act of deception itself.

References