Reliability and Validity in Selection Tests

Test reliability

The way of consistently a measuring the test characteristic is called reliability. If once again a person takes the test, can they get a similar test score, or get the different score. A test that defines similar scores for individual that once again tries the test is referred to measure a characteristic reliably. How can we define a person who does not have exactly the same test score most of the time he or she takes the test? The answers are Test taker's temporary psychological or physical state. Test performance is determined by a people psychological or physical state when testing. For instance, various levels of anxiety, fatigue, or motivation may affect the candidate test results. Environmental factors differ in the testing surrounding, like room temperature, lighting, noise, or even the test administrator, can influence a person test performance. Test form. Different test has different version or form. Item varies on every form, but each form determines the same thing. Various forms of a test are told as parallel forms or alternate forms. They have developed the similar measurement of characteristics, but they contain UN matching item, since the forms are not the same with one another, so the candidate might do better in one form than on another form. Multiple raters, in certain tests, score is measured on the basis of rater's judgments of the performance of candidates or responses. The training, experience, and frame of reference among raters help in producing various test scores for the test taker. The reliability of a test measures the reliability of coefficient. It is denoted by "r," and include number that range between 0 and 1.00, where r = 0 with no reliability, and r = 1.00 which indicate perfect reliability. It is difficult to find a test with perfect reliability. Generally, the reliability of a test is measured in decimal, for example, r = .90 or r = .73. Table.1 shows the guidelines that interpret test reliability. However, never select or reject a test solely on the basis of a size of its reliability coefficient. To define a test's reliability, we must consider the method of test, the method of reliability estimate reported, and the context in which the test will be used.

Table 1 General Guidelines for reliability

Test validity

The essential aspect in selecting a test is called validity. Validity defines the characteristic that test measures and tells the way in which the test is measured effectively. Validity defines whether the characteristic of the test is measured is related to job qualifications and needs. Validity shows the method and provides a concept to the test scores. The test validity is the evidence which estimates whether there is a relation between the performance of test and job. It defines what you can conclude or predict about another score on the test. If a test is used for a valid predictor of job performance, you can conclude that the individual is scoring better on the test and can perform goods on the job than persons who score low on the test other things remaining the same. Validity defines the way in which we can conclude or predict about people on the basis of their test scores. In other words, the usefulness of the test is validity.Validity indicates whether a test is measuring what it is supposed to be measuring such as does the test actually measure what it is intended to measure?

Methods for conducting validation studies

Criterion-related validation defines the correlation or statistical relationship between performance of test and job. In other words, people scoring on the test for performing a better job than people who score low on the test is called criterion. If the criterion is determined at the time where the test is given, it is also known as concurrent validity; if the criterion is gained delayed, it is known as predictive validity. Content-related validation defines that the content of the test which shows the essential behaviour of the job. In other words, test item must not measure indirectly the essential requirements and qualifications for the job. Construct-related validation defines that the measures of tests help in constructing or characteristic that it claims in measuring, and the characteristic is essential for successful job performance. The three types of validity-criterion-related, content, and construct help in providing validation support depending on the environment. They tend in overlapping, and, depends on the situation that is more appropriate.
The criterion-related validity of a test uses the validity coefficient. It includes a number from 0 to 1.00 which shows the relationship, "r," between the test and a measure the performance of the job. The greater the validity coefficient, the higher or larger confidence predictions can be made from the test scores. However, it requires more than one to predict job performance because the job success depends on different variable factors. Therefore, validity coefficients exceed r = .40.

Table 2 General Guidelines for Interpreting Validity Coefficients

As a general rule, the higher the validity coefficient it will be more profit in using the test. Validity coefficients such as r=.21 to r=.35 are useful for using an only one test. Validities for selection systems includes different tests that will probably be greater because we tend to use different tools for measuring various aspects of performance, where a one test tends in measuring or predicting some aspects of the total performance. Table 3 contains general guidelines that interpret test validity for a single test. To evaluate the validity of test we might require the services of a testing expert. In addition to the magnitude of the validity coefficient, we must also consider the minimum the factors that are given below:

• the level of adverse aspects that are associated with your assessment tool
• selection ratio such as the number of applicants versus the number of openings
• cost for hiring error
• the cost of the selection tool
• The probability of hiring a qualified applicant based on chance alone.

References

Dessler, Gary.a Framework for Human Resource Management.Florida: Pearson Education Asia, 2002.

http://www.hr-guide.com/data/G362.htm