Establishing Controls in Psychological Experiments | Experiments | Psychology

Some of the important methods of establishing controls in psychological experiments are: 1. Elimination 2. Matching 3. Counter Balancing.

Method # 1. Elimination:

Elimination is probably the most sensible and direct mode of establishing control in psychological experiments. Let us imagine a psychologist who wants to compare the abilities of men and women to detect an object which moves towards them from a distance. Here he finds that the average ability of men is less than that of women.

This result is surprising. On checking up his subject again he finds that some of the men suffer from short sight. Now he repeats the experiment after eliminating these men. This time he finds no difference between men and women.

Elimination is also done with reference to the experimental conditions. Imagine a psychologist who carries out an experiment to compare boys and girls on their ability to learn a puzzle. One group works in the classroom while the other group works in the library room. On their first performance the former group is found to be slower in learning.

However, later, the experimenter finds that the group working in the classroom was disturbed by the noises going on around. This experiment is now repeated after eliminating the noises and the results do not show any difference between the two groups. Here we see that in the first-instance subjects are eliminated and in the second an environmental condition is eliminated.

Method # 2. Matching:

It is not always possible, however, to employ the method of elimination. In many instances, we may not get enough individuals, if we go on eliminating. In such instances resort is made to another technique called matching. This refers to the selection of the subjects for the experiment. Let us take for example an experiment where an attempt is made to compare the general knowledge of boys with girls.

Here, in selecting the groups the psychologist should be careful. For example, if out of the 1000 girls he is studying, 500 are from rich class, 300 are from middle class and 200 from poor class, and again 600 of these girls are from a highly educated background, 300 from fairly educated background and 100 from uneducated background, he has to select the boys group also in the same proportions.

Let us imagine that this is not the situation and he selects 1000 boys, of whom 300 are rich, 300 are from middle class and 400 from poor homes; again 300 from highly educated, 250 from fairy educated and 450 from uneducated backgrounds. In such a case he may find a higher general knowledge in the case of girls. But we can immediately say that this is not a true difference between the boys and girls.

The average score of the girls may be higher because there is a distinctly higher proportion of girls’ coming from higher economic and educated homes. It is quite possible that in richer and more educated homes there are more facilities; books, magazines, parental teaching etc. Now, these benefits are available to a large number of girls compared to boys.

Thus, it follows that the difference between the boys and the girls is more due to these influences than due to any basic difference between the sexes. Now if the group of boys is selected in the same proportions of different economic and educational backgrounds as in the girls’ group the differences may not be seen any more.

The above example is an illustration of the need for matching. In fact matching of subjects is necessary in all psychological experiments. It is a very complex process. We have shown in the example how the two groups had to be matched on many more factors.

The exact factors are to be decided on the basis of previous research, as to what are the different independent variables, apart from the one studied in the experiment that can affect the response or dependent variables to be studied. Many of these may prove to be of a subjective nature.

Method # 3. Counter Balancing:

This is a third mode of establishing control and involves mainly a manipulation of experimental conditions. We have seen how some control on the experimental conditions can be enforced by the process of elimination. But it is not always possible to eliminate sources of errors. Here the method of counter-balancing or balancing is employed. The classical experiment on Muller-Lyer illusion in psychology is a fitting example here.

A is fixed in length and B can be varied by a sliding arrangement from very short to a large length. In this experiment, the experimenter goes on varying B until the subjects report that B is equal to A. This can be done in two ways. The experimenter may start with a very small length of B and gradually increase it.

This involves an outward movement of the eyes called condition ‘O’. If on the other hand he starts with a long B and goes on reducing it then this involves inward movements called condition T. Now because of this difference in the type of eye movements we often find the subjects making different responses under the two conditions.

The best result will be got by repeating the experiment a number of times, half the time under condition O and the other half under condition I. If only one condition is maintained the results may be wrong. Outward movement will involve greater strain on the eyes. As a result even short increases in B may be felt as longer due to the strain.

Similarly if B is always kept on the left of A, then those people who can see better with the left eye will be at an advantage over those who do not have this advantage (condition L). Similarly, keeping B always to the right of A (condition R) will give an advantage to people with better right vision. Hence, in addition to balancing conditions ‘L’ and ‘O’ as shown above, conditions ‘R’ and T must also be balanced.

The total number of observations must be equally distributed under the following categories:

RI, RO, LI, LO.

The need for matching and counter-balancing arises in all psychological experiments. To achieve this, psychologists follow certain well-established designs of experiments for advanced research. These are set patterns like building plans which can be adopted by an experimenter to choose his subjects and arrange his own experimental conditions. These designs are also given patent names as Latin-square design. Randomized group design etc..