Model answer outline :
An experiment is a method of observation under controlled condition. This method enables us to understand the cause – effect relationship between variables. In a typical experiment, the hypothesized cause of an incident is manipulated to see its effect by keeping other potential influences under control. Triplett, for instance, examined the hypothesis that social condition influences one’s performance. He repeatedly examined how rate of response is influenced by the presence of other people while keeping other factors under control. Data collected by him confirmed this hypothesis.
The science of Methodology emphasizes that an experiment is nothing but its control. Greater control means more internally valid inferences about the cause – effect relationship.
But controlling variances are not that easy in the social – behavioural sciences due to practical and ethical reasons. We can’t deprive infants, for example, of maternal love to see its impacts on their emotional and social development. As an alternative, we try to understand the results by conducting expost facto researches where effect has already taken place and we move in the backward direction from “effect to cause”. This method however, lacks the control over individual differences called randomization.
Even in the case of normal experiments in the social science with human subjects, it is more difficult than physical science because
1) we can not control everythings including hidden individuals difference
2) human subjects think and thus could make different interpretations of the same experimental treatments
3) human beings are not only reactive but proactive also. Demand characteristics of the experiment may predispose subjects to respond on the expected lines.
4) social reactions like impression formation , by the subject of the experimentor, may add and impurify the findings.
Moreover, a social reality unlike a physical reality often has several, rather than, single causes. Poverty for example has multiple causal factors that interact to produce the result. It necessitates to use a multi factorial design of the experiment which is more difficult to conduct than a linear experiment.