This is attempted using a process called Factor Analysis. A pioneer in this field was Raymond Cattell (1965). Cattell believed that the personality could be divided into two kinds: Surface Traits and Source Traits. Surface Traits were those, which other people could see, the overt personality, but underlying these are Source Traits which create the basis of the personality. Although Surface Traits do reflect Source Traits, they vary from person to person. Cattell considered that there were sixteen major source traits which everyone possessed to a lesser or greater degree and from this he developed the personality test known as the 16PF which has been used to provide profiles in many areas of psychology. However the most famous trait theorist is Hans Eysenck.
Eysenck was influenced by the Behaviourist tradition. When studying personality however, you cannot watch a person’s behaviour all day to interpret their personality. Eysenck adopted the approach of sampling different parts of a person’s behaviour by asking them questions about their normal activities. These tests were developed as questionnaires and by analysing the results he was able to develop a theory of personality which was a world away from that of theorists like Freud. Eysenck used factor analysis to identify basic aspects of personality; he investigated 700 servicemen who were being treated for neurotic disorders. He eventually concluded that there were two major dimensions to the personality, Extroversion and Neuroticism. Hans Eysenck’s model of personality considered the four personality types outlined by the ancient Greeks. Eysenck eventually created a psychometric test, which formed the basis of his whole theory; this is known as the Eysenck Personality Inventory or EPI and is the test we used for this investigation.
Psychometric tests are fairly short and are time efficient and many, like the EPI can be given to a number of people at the same time. This makes them cost efficient and prevent money being wasted, especially in employment sectors where they are used to select candidates for training like in the Air Force. Psychometric tests also need to be standardised. There are three parts to this process, the first ensures that everyone involved has the same experience and understand that it is the contents of the test that produce the results. The second aspect is developing population norms, which are general tables allowing comparison between one persons scores and another’s, like our Raw Score Data Table. The third is a little more complicated; involving making sure that the outcome conforms to established standards, and that everyone involved fully comprehends the implications and is comfortable with it.
Then there are the questions of Reliability. A psychometric test is no use if the results are affected by outside influence, for example, if an extroversion test reflected the mood you were in it would not identify your underlying personality. So, the test needs to produce consistent, reliable results, in this case we use the test re-test method to assess reliability.
Cochrane (1974),used the EPI questionnaires, to test Eysenck’s theory; ‘the criminal is a neurotic extrovert.’
The EPI questionnaire comprises of items of a ‘yes/no’ variety. They are essentially intended as research tools (as opposed to diagnostic tools for use in clinical settings) and, as such, ‘they are regarded as acceptable, reliable and valid’ (Kline 1981, Shackleton and Fletcher, 1984).
We also need to ensure that the test measures what it is supposed to measure, bringing in the question of validity. There are four types of validity: Face Validity, Criterion Validity, Construct Validity and Ecological Validity. Face Validity, or ‘eyeball’ is the simplest, where a test looks as though it measures what it is supposed to. On the whole psychometric tests measure what they claim to; so they have what is called an acceptable validity, making them less prone to error than other types of assessment.
Aim
The aim was to investigate the reliability of Hans Eysenck’s EPI Test.
Alternative Hypothesis
The Alternative Hypothesis will be that there is a positive correlation between the E and N scores of the forms A and B in Eysenck’s Personality Inventory. This is a one tailed test, because the hypothesis is directional.
Null Hypothesis
The Null Hypothesis will be that there is not a positive correlation between the E and N scores of forms A and B of Eysenck’s Personality Inventory, and any such found relationship is down to chance factors in the operation.
Method
The correlation study was carried out because we were looking for a relationship between the E and N scores for Tests A and B of Hans Eysenck’s EPI Test. The Independent Variable and the Dependant Variable were not operationalised in this case because an experimental method was not being used. This investigation used test re-test reliability.
Design
A correlational study was used because it is quick, economical and cancels out individual differences.
Participants
Eighteen students from a sixth-form tutor group, were used for this test. The type of sample used was opportunity because it was quick, easy and could be carried out in the Access group classroom.
Materials
The materials used were: Pens, paper, calculator, ruler, rubber, and Hans Eysenck’s EPI Test.
Procedure
An experimenter issued copies of Hans Eysenck’s EPI Test to a group of Access students who were then asked to complete it by reading the standardised instructions on the front page. Further, they were asked on completion of the test to write down their thoughts about completing it, and make notes on the reverse.
The participants were then de-briefed in accordance with British Psychological Society (BPS) ethical guidelines. The experimenter ensured that all participants were completely free of stress and were entirely satisfied that the exercise was a test of the EPI and not of the participants.
A Raw Score Data Table was then completed which highlighted the E and N scores on Tests A and B of the EPI.
Finally it was the E scores that were used in this investigation in order to establish the reliability of the EPI.
Results
From the Raw Scores produced, a scatter graph was developed. This shows that there is evidence that the scores show a positive correlation. However, we cannot accept an ‘eyeball’ test that this result, as evidence, is significant. Therefore the data was treated to further statistical analysis.
Treatment of Results
A Spearman Rho Correlation Coefficient Test was chosen because the data was Ordinal and parametric assumptions were not met. We then issued a statement based on the Critical Table Value. Ranking E and N scores–using Spearman Rho Correlation Coefficient Test
A result table illustrating the outcome of Eysenck’s personality inventory, illustrating the ‘E’ scores for form A and B.
Σ d ² = 126
rs = 1- 6Σ d ²
n(n² - 1)
= 1- 6 x 126
18 (18² - 1)
= 1- 6 x 126
18 (324 -1)
= 1- 6 x 126
18 x 323
= 1- 756
5814
= 1- 0.130031
= 0.869969
The Spearman’s coefficient of rank correlation for the relationship between the ‘E’ scores of form A and B is 0.87 (2dp). This is a strong positive correlation.
A result table illustrating the outcome of Eysenck’s personality inventory, illustrating the ‘E’ scores for form A and B.
Σ d ² = 89
rs = 1- 6 Σ d ²
n(n² - 1)
= 1- 6 x 89
18 (18² - 1)
= 1- 6 x 89
18 (324 – 1)
= 1- 6 x 89
18 x 323
= 1- 534
5814
= 1- 0.0918473
= 0.9081527
The Spearman’s coefficient of rank correlation for the relationship between the ‘N’ scores of form A and B is 0.90 (2dp). This is a strong positive correlation.
Statement
The Critical Table Value (CTV) is 0.464 at 0.05 level and 0.622 at 0.01 level. With Number of people being 18 and the significance of the data being judged at a level of accuracy to 0.05 for a one tailed test, the calculated value were; for the ‘E’ score 0.87 and the ‘N’ score 0.90. In order to be significant the calculated value must equal or exceed the CTV. In this case it does so we can therefore accept the alternative hypothesis and reject the null hypothesis.
Discussion
Our results showed that there was a highly significant correlation between the scores on tests A and B of Hans Eysenck’s EPI test. It can be stated therefore that the EPI is a reliable psychometric test, a point that we set out to prove with our original hypothesis. However, the sample used was opportunity because we tested it on ourselves and the size was limited with only eighteen subjects. Ideally a test like this should involve around 25-30 candidates to give us the results we expect, despite this we still had a significant result. The subjects were naïve so did not have a choice as to whether or not they consented to the test and were not given the right to withdraw. They were however de-briefed on completion and were satisfied that the test was of the reliability of Hans Eysenck’s EPI and not of themselves.
There have been many criticisms of Eysenck’s approach to personality testing. One of these is the fact that his original study in itself was limited, and many psychologists feel that he did not obtain a good range of personality characteristics from ‘normal’ individuals. They state that this may have resulted in his theory being biased towards certain kinds of personality. Heim (1970) argued that personality is much more complicated than Eysenck seemed to be suggesting, and that his use of factor analysis could in itself over simplify the procedure.
Another general criticism is that the use of questionnaires could mean that the results are easily influenced by the moods of those participating. A question like ‘Would you rather stay at home on your own or go to a boring party?’ may be answered in different ways by the same person depending on how they felt at the time. This criticism was put forward by Michel (1968). Personality questionnaires may also produce response sets; i.e. stereotyped ways of responding. This may take the form of acquiescence, simply answering yes regardless of the question; also social desirability, giving the answers that you believe will present you in the best light. Some questionnaires, like the EPI are also derived from forced choice items, in the case of the EPI, yes or no answers. Meaningful and valid comparisons of personality cannot really be made from such forced choice questions. However, many of the best questionnaires have a great deal of evidence to support their predictive validity, or their ability to predict a person’s performance in some related areas of functioning.
Limitations of design
The main limitations of the method in this investigation are that a correlation study establishes a relationship, but not a cause and effect relationship. The only way to be certain about this result would have been to devise an experiment to test the reliability of the EPI. However, the investigators failed to achieve this at the design stage.
Design Problems
The design used was a repeated measure and problems with this could include order effects such as learning, fatigue or boredom and subjects given the same test may do better due to practice. Also demand characteristics may become a problem; as the subject completes both conditions of the research, they may guess the aim of the investigation and act differently.
Conclusion
It has already been stated that this investigation was a one tailed test, with the number of people being eighteen. The calculated value for the ‘E’ score was 0.87. In order to be significant it had to exceed the critical table value, and in this case it did, at 0.01 level. Therefore we accepted the Alternative Hypothesis that there would be a positive correlation between the ‘E’ scores for forms A and B, and we rejected the Null .
We also used this comparison for the ‘N’ score. With the number of people being eighteen. The calculated value for the ‘N’ score was 0.90. In order to be significant it had to exceed the critical table value, and in this case it did, at 0.01 level. Therefore we accepted the Alternative Hypothesis that there would be a positive correlation between the ‘N’ scores for forms A and B, and we rejected the Null .