Since there are many facets of ability, interest, personality, attitude, the interests of psychologists are many faceted too. Different psychologists have developed various tests in order to measure many of these facets. Developing of such tests is the most significant factor of most psychological testing. The tests selected are mostly targeted to measure the special abilities on personality attributes of interest. It is also checked whether the test selected is completely sound, and that it will yield constructive information of real value. There are many ways in which it can be ascertained whether the test is sound or not.
Although there are no such techniques which can confirm the soundness of a test, but there are few important characteristics that define a strong test. Most of the organisations look for tests which are extremely trustworthy and informative. Reliability, validity and norms are the three important qualities which any test must have in order to be standardized and strong. Let us look at the characteristic features of good standardized test.
For a test to be standardized and well acceptable it must be reliable. This means that the results obtained by the test at a certain point of time, must yield the same result when tested at a different point of time. A test that yields different results as time varies, cannot be called a reliable test. So in order for a test to be reliable it must produce similar results even though different testers administer it, different peoples score it, different forms of the tests are given, and the same person takes the test at two or more different times. The different sets of scores obtained by the same individual, at different points of time, when given the same test is usually checked by comparing the different sets of scores.
The test retest reliability criteria of a test is successfully fulfilled when a test gets similar results when it is administered twice on the same individual, at different points of time.A high test-retest correlation or a high test retest reliability quotient indicates that the test is extremely reliable.But in actual practice, psychometric testcannot be made hundred percent reliable and that is absolutely normal.One reason might be that human beings are dynamic, and that meaningful changes do occur inindividuals over the time. Like for example, if any candidate who scores comparatively low in his group at an initial testing may develop new skills that elevates his position in the group, by the time of the second testing. Also it might so happen that he may have practised or learned the skill, and acquired it better. When the correlation coefficient of Reliability false below .80 it is considered to be low reliability.
Almost all personality tests have lower reliability and did this may be due to partly of the instability of the things like attitudes and feelings which personality tests are designed to measure. Alsotraits of a person can change over time, and it might slightly very in different situations also.Hence personality correlation coefficients are natural to have low correlation coefficients. In order to adjust in different situations, human beings acquire different characteristics. These characteristics or traits accumulate together to form personality. Correlationcoefficient can range from 0.0 to 1.0. Generally the reliability coefficient of correlation is obtained around 0.90 or more in intelligence tests. 1.00 indicates a perfect correspondence and 0.0 indicates no correspondence at all.
Low reliability usually means that the test in question has a high probability of error which means that any predictions we make about people based on such test will be risky. Depending on the quantity of error each test incorporates, the hypothesis propounded by the test can be accepted or rejected. If tests with low reliability are used at all, their scores should be interpreted with caution. Also it would be better if we take help of other tests in such a case.
In order to improve the figures on the test one can administer the test more carefully. If the tests are not administered and scored by a truly standard procedure, there always remains a possibility of error. Testers who are not well aware of the testing procedures, which put the test instructions in their own words, may be preventing the test experience from being uniform across all the testing.Standardized instructions should be used in order to instruct the candidates. Instructions should be such that they are well understood, said or written in a lucid language, clear and district.
Situations can be controlled as such candidates might be asked to not take any extra preparations after they have been exposed to the first instance of testing. They must be encouraged saying that it will yield a more accurate and better result, and will eventually help in a clear assessment of them.
Also, the test administrators may be giving a somewhat different version of test each time they administer it. And highly variable, unreliable sources may have resulted. Thus bymaking the test procedures uniform we can make the tests more reliable.
Apart from test retest reliability, there are other forms of reliability like, inter rater reliability, internal consistency reliability, parallel forms reliability, etc. Split half reliability and average inter term correlation are the two different branches of internal consistency reliability.
Hence, reliability is a strong technique of which determines the extent to which a test can produce accurate and consistent results.
Author's Score 1.7
Up Votes 3
Down Votes 0
Voted on 2 articles
Comments on What Do We Expect Of The Reliability Of Psychometric Tests?