The research subjects, Milgram emphasized, were ordinary people--typical citizens from the professional and working classes. And they obeyed even when no further reason was given except these statements: "The experiment requires that you go on. . . . It is absolutely essential that we continue.''

On publication, Milgram's findings generated a wide range of commentary. Many readers thought about the astonishing obedience of the Nazis in carrying out brutal, inhumane orders. More than one critic decided that Milgram had accomplished some of the most morally significant investigations in modern psychology (Elms, 1972). Another described Milgram's work as "a momentous and meaningful contribution to our knowledge of human behavior'' (Erikson, 1968). In sharp contrast, another decided that Milgram himself had behaved immorally, duping his subjects and persuading them to perform this distasteful task. No matter what they revealed, his findings were not worth the tension and self-doubt they created in his subjects (Baumrind, 1964). Many others have voiced opinions between these extremes or contributed different interpretations of his findings (Nissani, 1990).

These diverse views offer an opportunity to examine research methods from a broader perspective, focusing on ethical issues and comparisons among the different methods. Depending on the situation, each method has special assets and limitations.

This question of research ethics concerns the experience of subjects participating in a scientific investigation. It requires the humane and just treatment of all subjects. This issue is regularly raised in animal research by psychologists and the lay public (Timberlake, 1993; Ulrich, 1991). It is also confronted in the clinic, usually by the investigators themselves (Hall, 1991a).

People commuting to work may become subjects in naturalistic observation, but they do not alter their behavior in any way. Ethics should not become an issue, provided that the results are managed with discretion and confidentiality. People mailing lost letters go out of their way, but they do so voluntarily, just as they do to assist an elderly person, use a trash can, or help someone with parking. The general rule is that the subject's explicit consent to participate is not required if the behavior in question occurs as part of a normal routine, if there is no coercion, and if the findings are handled in a manner that fully respects the right to privacy.

Basic Ethical Questions. When Jack Washington participated in Milgram's experiment, he went out of his way, was duped by a rigged drawing for the teacher and learner roles, and received false information about the electrical shocks. These are more fundamental ethical issues. Such issues are not unique to psychology; they occur in all disciplines. For some people, the entire field of nuclear physics is an ethical issue. In biology, animals are confined to cages and subjected to surgery solely for research purposes. In legal and psychiatric research, ethical issues arise over the question of clients' rights. Even educational research involves ethical considerations, for the procedures sometimes impose researchers' values on the students in the study (Figure 2­9).

Figure 2-9 Research Ethics. Issues of moral values and moral behavior underlie all research. When research subjects are exposed to a prison environment, for example, questions immediately arise concerning their physical and mental welfare.

The issues in animal rights, for example, are more complex than might be readily apparent, as demonstrated when a cat lover complained that her neighbor's pet boa constrictor devoured kittens. The neighbor replied that the snake merely ate mice and pointed out an advantage of snakes as pets. Having extraordinarily low metabolism, they consume far fewer fellow creatures than the approximately 54 million pet cats in this country. One reason for the cat lover's complaint, of course, is that cats rank higher than snakes on the human list of pet preferences. In killing and preparing cats' food, human beings make attractive packages out of the dead fish, horses, rats, and other animals fed to felines. In short, the animal world pays a high price for our pets, which we keep chiefly for one purpose--affection (Herzog, 1991). And the proper relation of human beings to lower animals, whether in research or relaxation, remains a perplexing ethical issue (Locke, 1992).

One basic ethical question in research seems to be this: Does any possible discomfort incurred by all subjects collectively outweigh the gain in alleviation of human and nonhuman problems? Many of Milgram's subjects experienced tension, doubt, and self-recrimination, clearly signs of discomfort. On the other side of the ledger, the obedience of the Nazis played an important role in the pain experienced by millions of people in World War II, and after their capture they did not appear to be cruel and sadistic. Instead, they seemed like ordinary citizens who did what they thought was their duty, which is the way many of Milgram's subjects characterized themselves. After the experiment, they decided that they had to resist authority more effectively in the future. Thus, the question still stands: On balance, was this research justifiable? Was the suffering of Milgram's subjects offset by the gain in understanding obedience to malevolent authority?

This ethical question involves value judgments about the feelings of research subjects. It also involves value judgments about the role of research in society. It extends beyond obedience and disobedience to include legitimate versus illegitimate authority and even benevolent versus malevolent orders (Saks, 1992). Taking a thoughtful ethical stand is a challenging task.

Among these, the most important is informed consent, which means that the general nature, risks, and benefits of research participation are explained to each subject before the procedures begin. The subjects are told about the tasks they will be expected to perform, and if they agree to participate, they do so giving informed consent. In Milgram's work, all subjects were informed of the basic experimental procedures at the outset. Each of them signed a consent form, indicating voluntary participation in this research.

An obvious question in informed consent concerns the complexity of the explanation and, in some cases, the need for deception about underlying purposes. Sometimes a less-than-full explanation may be acceptable, providing the subject incurs no other risk by participation. Still another issue is the subject's ability to give informed consent, especially in three populations: elderly people, the mentally disabled, and children. Investigators studying young children typically rely on a proxy from their parents, but each research project, with its unique characteristics, requires special steps to ensure that children's rights are respected (Hughes & Helling, 1991; Krener & Mancina, 1994).

For these reasons, an interdisciplinary ethics committee now must evaluate all federally funded psychological research. The goal is to permit the most useful, productive research without endangering the subjects in any way. The ideal solution to this problem, and one toward which many psychologists are working, is the development of research techniques that rely on the subjects' usual behavior and naturally occurring events rather than on contrived situations and misleading information (Huntingford, 1984; Kelman, 1967).

Debriefing Interview. At the close of each research session, Milgram interviewed the subject. This procedure is called a debriefing interview, and it has a dual purpose: to ensure the maximum benefit for the subject and to obtain any further information that might be useful. Concern for the subjects is paramount, directed not only to ensuring that they are as fit and healthy as they were before the research but also to enable them to profit as much as possible from the experience.

Milgram's debriefing instructions included a meeting between the teacher and the unharmed learner, as well as extended discussion with the authority figure. During his debriefing interview, Jack Washington explained that he would have discontinued the shocks if he had received a signal from the experimenter. "I did not get a cue to stop,'' he emphasized. Then he added that he knew the shocks were painful but not dangerous. Earlier in life he had received a very strong, accidental shock to his hand, causing him pain but no permanent damage, and thus he believed the investigator's contention that the shocks were not harmful.

Some weeks later, all subjects received a five-page report, including a questionnaire asking them to express once again their thoughts and feelings about the experience. Altogether, 80% of the subjects recommended more experiments of this nature, and 74% replied that they had learned something important to them personally. For Milgram, this finding constituted the central justification of this research: The participants judged it as acceptable and worthwhile (Milgram, 1974).

This work is accomplished by all four methods, combined and modified in various ways to fit the requirements of a particular research question. We know about Milgram's use of these methods through his collected reports, The Individual in a Social World (1992), and his book with a single focus, Obedience to Authority (1974). In response to continued interest 15 years after publication of the former volume, two of Stanley Milgram's followers, John Sabini and Maury Silver, edited and republished that book as a second edition under his name (Milgram, 1992). These works show how Milgram found ways to collect research information by using an original mix of methods: observation, the survey, the case study, and, of course, laboratory experiments (Zimbardo, 1992).

Statistical Analyses. After collecting the information, the investigator's next task is to analyze and interpret it. Sometimes qualitative analysis is involved, requiring considerable wisdom on the part of the investigator. Especially in naturalistic observation and case studies, subjective judgments and verbal descriptions may be the only recourse for depicting the findings. More often, quantitative analysis is included, meaning that the information is expressed in numerical units and statistical techniques are employed to assess them. These techniques, discussed in detail in the final chapter, can be categorized as descriptive, correlational, and inferential statistics.

Utilized in all research methods, descriptive statistics characterize or summarize a group of scores, often by presenting just a few numerical values. Usually they include some typical score, such as an average, and some measure of variability, indicating the extent to which other scores differ from the typical score. For example, Milgram found that commuters in New York City had an average of 4.5 familiar strangers in their lives. In one instance, the range was from zero, for a commuter new at the depot, to more than 12 for long-time commuters who recognized almost every "regular'' at the station. These numerical values are descriptive statistics.

Instead, an investigator may want to know about the relationship between two sets of scores. In his studies of pedestrians, Milgram noted that the size of crowds varied considerably according to several factors. We might wonder about the relationship between pedestrians and sidewalk performers. When pedestrians increase or decrease in number, do sidewalk performers also increase or decrease? Answers to such questions involve correlational statistics, which indicate the association between two sets of scores. They show whether there is a relationship between the number of pedestrians and the number of sidewalk performers.

Numerically, these relationships can vary from 0.00, meaning no correlation, to +1.00, a perfect positive correlation, or to -1.00, a perfect negative correlation. A correlation of +.60, for example, indicates a direct, rather strong association. As pedestrians increase in number, so do sidewalk performers. A correlation of -.10 indicates a weak, indirect association. As the number of pedestrians increases, the number of sidewalk performers tends to decrease mildly. However, correlational statistics do not indicate causality. Both factors may be influenced by a third variable, such as the weather or a holiday season.

Finally, inferential statistics are used to make a guess, an inference, or a statement of probability about one or several groups of scores. In Milgram's studies of obedience to authority, the subjects were asked to use a 14-point scale to estimate the amount of pain experienced by the learner. For a sample of obedient subjects, the mean was 11.36; for defiant subjects, it was 11.80. Is there a statistically significant difference between these two means? Expressed differently, what might be found with other samples of obedient and defiant subjects? Clearly, Milgram could not test all possible subjects. Inferential statistics allow the investigator to estimate the probability that the same findings would occur if the same experiment were repeated over and over again. Inferential statistics are common in the experimental and other research methods as the investigator tries to determine the reliability of the findings.

To emphasize the basic point once again: Research methods are used to collect information. Diverse statistical methods are used afterward to analyze and interpret this information.

Continuum of Methods. When these research methods are compared, they fall to some degree on a continuum of intervention or control. That is, they differ in the extent to which the investigator intervenes in the subjects' lives.

In naturalistic observation, at one end of this continuum, the subjects are studied in their own environment, unaffected by the investigator insofar as possible. In the survey, further along the continuum, the subjects' privacy is invaded to some degree by mail or telephone. Toward the opposite end of the continuum, case studies often take place in a clinic or comparable institution. At the far end of this continuum, opposite from naturalistic observation, the experimental method involves the fullest control and the most intervention. The subject usually enters a laboratory setting, and the investigator manipulates this environment in varying degrees, permitting the most definitive study of cause-and-effect relations.

There are innumerable variations in these methods. A study of animal migration might involve a combination of naturalistic observation and the survey method. Experiments can be conducted in school, at work, and elsewhere in the natural environment. Called field experiments, they reflect modifications in the basic experimental design. The overall aim is to find the most useful integration of methods for any topic of inquiry (Banaji & Crowder, 1989; Conway, 1991).

Collectively, these different methods provide a variety of useful approaches to the study of psychological problems. Each has advantages and disadvantages, and each makes a special contribution (Table 2­7). As these diverse methods become more fully developed and integrated, through psychologists' continual search for improved research techniques, we should become more and more skillful in understanding all sorts of behavior, including conformity, compliance, and obedience.

When Milgram performed these experiments with women, he found that the level of obedience was virtually identical to that of men. When these experiments were repeated by other investigators in Italy, Australia, Germany, and South Africa, each time with a somewhat different sample of subjects, the level of obedience was as high as that found by Milgram. In one instance, the subjects were required to administer psychological punishment, rather than physical punishment, harassing and berating an interviewee. Once again, the level of obedience was comparable to that in Milgram's work (Meeus & Raaijmakers, 1986).

These replications have essentially verified Milgram's findings on obedience to authority. Scientists within and outside of psychology therefore regard this work as a rare integration of the humanistic and empirical perspectives and as an extraordinary contribution to the study of moral issues. Panels of experts have selected this research as one of the most important investigations in modern psychology. With the exception of Darwin's book, it was the only single piece of research included in a comprehensive list of terms and concepts depicting basic psychological information (Boneau, 1990).

Milgram showed that what subjects say on a questionnaire may be very different from what they actually do. Further, he showed that social situations can powerfully override personal dispositions in influencing behavior. This conclusion is widely accepted, although personal factors cannot be entirely discounted (Blass, 1991). Finally, he showed that what people do is not necessarily predictable through common sense (Milgram, 1974).

Sometimes common sense is supported by psychological research. These occasions are gratifying, indicating that we learn something about human behavior and experience through daily life. Sometimes common sense is contradicted by psychological research. These occasions are important because scientific research is generally considered the most rigorous and reliable pathway to knowledge. A psychology that consistently opposed common sense would be disturbing, apparently concerned with a different reality than the one we think we know. A psychology that merely supported common sense would be a waste of time. The value of employing diverse research methods in modern psychology is that they increase our chances of distinguishing one from the other.

For his simple, elegant research methods, Stanley Milgram became internationally famous. For his studies of obedience, he was awarded numerous honors, including the prize in socio-psychology offered by the American Association for the Advancement of Science. This series of experiments has been described as one of the few classic studies in social psychology, containing all the elements of a parable: a story line, conceptual simplicity, vivid imagery, and an unexpected outcome (Kotre, 1992). Stanley Milgram completed these obedience studies while he was still in his twenties. Sadly, this early success was followed by an untimely death many years before it might have been expected. His genius is now to be found frozen in his published works (Sabin & Silver, 1992). He has left us an inspirational legacy in research methods, morally significant findings, and a reminder about our own lives: "We are all fragile creatures, entwined in a cobweb of social constraints."