Psychological Errors

An experiment in psychology has three types of variables viz. independent, dependent and intervening variables which we have been dealt with in lesson 3. The independent (stimulus) variable is manipulated or varied to observe its impact on the dependent variable (response). The dependent variable is the change in response or behaviour observed in the subject following or accompanying the changes in the independent variable. In an experiment we study the relationship or connection between the independent and dependent variables. These two variables stimulus and response are observable and measurable. The intervening variable which also exists and influences the results of the experiment in not the focus of attention in the experiment since it is, not seen being present within the subject but it distorts the true relationship between the independent and dependent variables. It is therefore, essential to take account of the intervening variables and control their effects. Failure to control their effects will introduce various errors in the experiment.

Sources of Errors

In any experiment, there are different possible errors that emanate from different sources and also are of different kinds. The possibility of errors in a psychological experiment is much greater because we deal with human subjects and also because of the very nature of the processes being experimented. There are three main sources of errors: 

1. some errors are a result of factors in the subject participating in the experiment 
2. some errors may result from the defects in the presentation of the stimulus
3. a few errors arise from the manner of recording or measuring the response. 

In a psychological experiment the subjects taking part in an experiment vary among themselves in many variables. There are individual differences in abilities, intelligence, motivation, learning, personality etc. These differences are bound to influence the results of the experiment in a predictable and systematic manner for e.g.in an experiment on the effectiveness of two teaching methods the subjects in the 2 groups cannot be of the same level of intelligence, motivation, learning, comprehensive power etc. but the subjects are bound to be different in these dimensions which will influence their learning and therefore will influence the experimental findings. This influence will change from trial to trial or even day to day. The errors of this type which remain stable, act in a particular direction and are systematic are called constant errors. 

There are also other types of errors which result from momentary variations in the subjects for e.g. students may be tired one day and not the next day, some internal conditions in the subjects like hunger, thirst, fatigue etc. which vary and do not remain stable may also introduce certain errors. Unlike the constant errors they do not remain stable and do not necessarily operate in the same direction. Such errors are called variable errors. In an psychological experiment" both constant and variable errors occur due to the characteristics and conditions of the subjects. IV is easy to estimate and control constant errors through careful selection of subjects. The variable errors are difficult to be eliminated completely. 

Errors resulting from experimental conditions 

In psychological experiments it is not only difficult to eliminate variations among subjects but it is also difficult to maintain identical experimental conditions for all the subjects on all the occasions. In the experiment on the effectiveness of the two teaching methods it is impossible to completely equate the two sets of teachers on their teaching ability, knowledge level etc. and if the experiment is carried out on different days and with different groups of subjects, it is not possible for the teachers to maintain the same level of teaching on all the occasions. Further, it would become much more difficult if a large number of teachers are involved. Similarly, there may be variations in the classroom atmosphere and other aspects. Such variations in the experimental conditions can influence the findings, resulting in errors. These errors may be constant errors or variable errors. For instance, the teacher using the discussion method may have a more effective personality that the teacher using the lecture method. This will result in constant error in favour of the discussion method. On the other hand, variations in the class room atmosphere can produce variable errors. 

Types of Errors

It is apparent from the above discussion that the psychological experiments can be contaminated by errors. When these errors creep in, the reliability, validity and accuracy of the experiment are adversely affected. Errors in psychological measures are of 2 types: Constant Errors and Variable Errors.

Constant Errors are those that occur repeatedly and are constant. They are 'also called systematic errors. 

Variable errors results from factors as the subject's mood, motivation. Then also called random errors because they occur randomly and not in a particular direction. 

The constant errors can be of 2 types Type 'S' & Type 'G' errors. The constant errors arising out of the variations among the subjects are called Type 'S' Errors. 

The constant errors resulting from stable differences in the experimental conditions are called Type 'G' errors. The variable errors resulting from both variations in the subjects and variations in the experimental conditions are called Type 'R' errors. Such errors are also known as 'sampling errors'. 

We need to be aware of the presence of the various types of errors in the experiment. These errors have to be minimized as far as possible by adopting proper methods of experimental control. 

Summary

In a psychological experiment, there are different types of errors likely to occur that emanate from different sources and are of different types. these errors may result from certain factors in the subject as well as the experimenter, from the defects in the measuring conditions and defects in the measuring tools. These are broadly classified into constant errors and variable errors. The constant errors may further grouped into type 'S' and type 'G' errors. The variable errors resulting from both the above sources,are called type 'R' errors or sampling errors. These errors should be minimized by establishing methods of experimental control. This will result in accuracy, reliability, and validity of the experiment.