This section draws on material from 'Research Methods' by Patrick McNeill (Routledge 1990)
The logic of the scientific method was set out by John Stuart Mill in 1843, and was named the method of difference. A simple example of what he meant by this is to take two glasses of water which are identical in every respect. Introduce a few drops of ink into one of these glasses. The water changes colour! According to Mill's method of difference it is safe to assume that the change in the colour of the water is due to the introduction of a new factor - the independent variable - in this case, the ink.
This example illustrates how scientists try to manipulate the causative factors which are seen as important (the independent variables), in order to observe and measure their effect on the subject under study - the dependent variable (in this case, the colour of the water).
In a laboratory experiment, the subject of the experiment (for example, water, or a chemical compound) is divided into two identical groups, which are maintained under strictly controlled conditions. One group - the experimental group - is then exposed to the independent variable. The other group, the control group, is not changed in any way but held in the controlled situation so as to eliminate those factors which are not the subject of the experiment, but which could contaminate it. Thus the experimenter is able to ensure that changes in the experimental group are the direct result of the introduction of a particular independent variable.
Some early sociologists believed that social phenomena were no different from natural phenomena. Therefore, social scientists assumed that they could use the research techniques of the natural sciences. However, because of the fundamental differences between inanimate objects and human subjects, sociologists have experienced difficulty in trying to use the experiment as a tool of social research. It is possible to illustrate some of the problems involved in such research by using examples drawn from social psychology, well known for its expertise in drawing habits out of rats...
Research by the social psychologist EIton Mayo at the Hawthorne plant of the Western Electric Company in 1925 provides a useful illustration of one of the major problems for social science in using experiments - the experimental effect.
The research at the Hawthorne plant was started by the factory management in order to ascertain the effect of various changes in working conditions on industrial output. This information in itself should alert you to the fact that social research is not intrinsically radical/left wing - call it what you will. This was research designed to increase profits for employers.
Mayo found that manipulating various independent variables such as lighting, the repetitive nature of the workload, the pace of work, and the spacing of rest periods, could not explain changes in workers' output. In particular, even when conditions were made less favourable to the workers, output still increased. Mayo's conclusion was that social factors were involved.
By trying to establish an experimental situation in which all factors were controlled, the researchers had in fact introduced into the situation a new factor, the social influence of the research process itself. Variation in performance was not linked to physical changes, but to the amount of attention the workers received. This finding is now known as The Hawthorne Effect.
In addition to the effect noted above, there was one other serious problem with Mayo's research design - the use of volunteers. Volunteers are by their nature likely to be unrepresentative.
A major difficulty in research using humans concerns ethics. Of course there are ethical considerations to be made in experiments on other animals, but humans present unique problems. For example, if participants are aware that they are part of an experiment then they may well alter their behaviour. One way to avoid this problem is not to reveal to participants that a particular experiment is occurring, but is this ethical? It is not, of course, only social scientists who are faced by this problem; natural scientists have been involved in fraudulent behaviour as regards the effects of radiation and drugs where the participants in the research were not kept well informed.
Some research will always be seen as unjustified, and certainly any research that can be seen as doing people harm. A controversial experiment published in 1968 by Rosenthal and Jacobson illustrates this problem. The aim of the experiment was to test the hypothesis that poor children perform badly at school because of teacher expectations, and not because they are members of disadvantaged groups.
In a school in San Francisco, teachers were asked to administer IQ tests to their pupils. The teachers were told that the tests were designed to predict intellectual gains in children. The teachers, who were unaware of the experiment, were then informed that 20% of the children were expected to show unusual intellectual gains in the year ahead. In fact, the 20% had been chosen by random selection and the difference between them and the other children, who constituted a control group, was entirely in the minds of the teachers. This, incidentally, is an excellent example of believing is seeing as opposed to seeing is believing, as a way of examining the understanding that people develop to explain situations.
The same tests were then administered to the children on three more occasions over a period of about 18 months. Results indicated that children from whom teachers expected greater intellectual gain did in fact exhibit such gain. Was this experiment fair to those children who were not chosen, but who could have benefited similarly from teacher expectations? This research is, of course, an excellent example of the self-fulfilling prophecy, and of W.I. Thomas's dictum, 'If men define situations as real...they are real in their consequences.'
It's important to note here that a further part of many ethical dilemmas is that information that could benefit some people might result from experiments that could harm other people, or at least deceive the original participant in the research.
The experiments of the social psychologist Stanley Milgram (1965) have also raised serious ethical consideration, in particular his research in obedience to authority. Milgram was interested in the effects of authority on human behaviour. He set up an experiment in laboratory conditions where paid volunteers were asked to co-operate in a learning experiment. On arrival at the lab, a second volunteer and the researcher met them. Lots were drawn to see who would be the teacher and who the learner. The learner was strapped into a chair - this was watched by the teacher. The teacher was then taken into another room and shown an apparatus that could deliver electric shocks to the learner. The teacher was then instructed by the researcher to teach the learner a series of linked words and to punish the learner with a shock if he gave a wrong answer.
The whole experiment was a set-up. The learner was in every case a co-worker of Milgram's. No shocks were actually transmitted, and the experiment rather than being about learning (which is what the volunteers were told) was about how far people would obey orders. In Milgram's defence it should be pointed out that much of the complaint about this experiment has focussed on the results and clearly, as Milgram himself has argued, you can't know the results in advance.
Another example of this type of research was reported by Haney, Banks and Zimbardo (1973). In this case, a simulated prison was created and 24 volunteers were assigned to the roles of guard or prisoner. After instruction in the regulations of the prison, the volunteers were placed in it for nearly a week. The guards were told that the research interest was the behaviour of the prisoners. The effects were dramatic. Five prisoners had to be released because of extreme emotional depression.
The experiment had to be ended early because of the behaviour it was inducing in both guards and prisoners. Only two prisoners refused the offer of early release in exchange for their fee as a volunteer; nearly all the guards expressed regret at having to finish the experiment early. Many of the guards showed that they enjoyed the power that they had.
Due to the difficulties of experimenting on humans, animals are sometimes used in psychology in the hope that they can reveal something that is applicable to human behaviour. For example, monkeys are relatively easy to control and experiments can be performed on them that would be considered unethical if humans were involved. However, it is questionable to what extent we can make assumptions about human behaviour on the basis of animal experimentation, and increasingly it is argued that animals have 'rights'.
Partly as a response to both practical and ethical criticism of the laboratory experiment, some researchers have conducted field experiments. These take place in a natural setting and those involved do not know that an experiment is being conducted. One example is that of Rosenhan who conducted research into the diagnosis of patients admitted to mental hospitals.
Another example is Sissons (1970). He arranged for an actor to dress up in a suit and bowler hat and stand in Paddington station, periodically asking people the way to Hyde Park. The same actor then changed into the clothes of a labourer and asked people the same question, with the same wording. Sissons studied the different reactions of those questioned. Sissons argued that any systematic variations must have been the result of the reactions of passers-by to the apparent social class of the person they were talking to, since all the other variables were held constant (but could this be true?).
A major problem for field experiments is that there is no way that the researcher can hope to control all the possible independent variables. The problem works both ways. If we control the situation we make it artificial, if we want the situation to be realistic, we have to accept that we cannot control all the variables.
A variety of field experiment used by ethnomethodology is the 'What if' experiment. The focus is on impromptu experiments that it is hoped will throw light on taken for granted situations and basic human interaction. Garfinkel has attempted to discover some of the basic features of interaction. His strategy is to start with a situation viewed as normal and then systematically disrupt it. In one experiment, students were asked to spend 15 minutes to an hour acting out the idea that they were boarders in their own homes. Garfinkel told them to act in a polite fashion, to use formal address, and to speak only when spoken to. In many cases, family members were stupefied and tried to restore the situation to normal. Reports were filled with accounts of astonishment, bewilderment, shock, anxiety, embarrassment and anger, and with charges by various family members that the student was mean, inconsiderate, selfish and impolite. In all cases, the situation was restored upon explanation, but for the most part, family members were not amused.
Experiments of a similar kind have been used with the object of discovering the way in which people use space to create territories. Robert Sommer, a psychologist interested in the influence of architectural design on behaviour, argued that the best way of learning the location of invisible boundaries is to keep walking until somebody complains.
Sommer used this invasion technique in a series of experiments. In a study area of a college library, observation over a two-year period showed that the first occupants of the room usually sat one to a table at end chairs. Sommer's female research assistant chose women who were sitting alone, with at least one book in front of them, and who were surrounded by empty chairs. She then violated library norms by invading their territory by sitting either directly opposite, or next to them. The subjects reacted in different ways - defensive gestures, shifts in posture, attempts to move away. Other experiments investigated the way students defended their territory, and whether people in an area for a long time established territorial rights.
Thus there are a number of ways in which the experiment can be used within the social sciences. The experiment is a useful tool in the discovery of the basic elements of interaction, but it is not a very valid way of testing complex social relationships.