the experimental setting. For example, the experiment may be conducted in a sound-proof and air- conditioned room to eliminate the effect of noise and temperature. Elimination is not always possible. In such cases, effort should be made to hold them constant so that their effect remains the same throughout the experiment.
For controlling organismic (e.g., fear, motivation) and background variables (such as rural/urban, caste, socio- economic status) matching is also used. In this procedure the relevant variables in the two groups are equated or are held constant by taking matched pairs across conditions of the experiment. Counter-balancing technique is used to minimise the sequence effect. Suppose there are two tasks to be given in an experiment.
Rather than giving the two tasks in the same sequence the experimenter may interchange the order of the tasks. Thus, half of the group may receive the tasks in the order of A and B while the other half in order of B and A or the same individual may be given the task in A, B, B, A order. Random assignment of participants to different groups eliminates any potential systematic differences between groups. The strength of a well-designed experiment is that it can provide, relatively speaking, a convincing evidence of a cause-effect relationship between two or more variables.
However, experiments are often conducted in a highly controlled laboratory situation. In this sense, they only simulate situations that exist in the outside world. They are frequently criticised for this reason. The experiments may produce results that do not generalise well, or apply to real situations.
In other words, they have low external validity . Another limitation of the laboratory experiment is that it is not always feasible to study a particular problem experimentally. For example, an experiment to study the effect of nutritional deficiency on intelligence level of children cannot be conducted as it would be ethically wrong to starve anyone. The third problem is that it is difficult to know and control all the relevant variables.
Field Experiments and Quasi Experiments If a researcher wants to have high generalisability or to conduct