9 - Selection Tests
When is a content validity study appropriate? When is it inappropriate?
A test can be supported by a content validity procedure to the extent that the test is a representative sample of job content. The content validity approach is not appropriate for validating tests that measure traits or constructs such as personality, intelligence, aptitude, etc. Nor is it a proper procedure when the test involves skills, abilities, or knowledge that the employee will be expected to acquire on the job.
Review the disadvantages and advantages of tests? In your opinion, do the advantages outweigh the disadvantages? Why?
Advantages: -Objectivity- results are not influenced by evaluator -Cost-effectiveness- inexpensive way to choose workers and improve organizational productivity -Quality of information- provide information other selection devices can't -Validity- provide quantitative data, therefore capable of objective statistical validation -Legally defensible-establish a demonstrable relationship between test scores and job performance Disadvantages: -do not measure motivational levels- can do vs will do -more accurate at predicating failures than successes -more predictive for groups than for specific individuals -Sometimes susceptible to dishonesty -create anxiety on the part of the test taker -subject to legal challenge.
What court cases spell out rules about legal uses of employment tests? What do these cases say is legal?
Albermarle v. Moody, the Court ruled that any test used in selection or promotion decisions must be validated if its use has had an adverse impact on protected classes. Additionally, the Uniform Guidelines promulgated standards to which tests must adhere if they are to be considered nondiscriminatory. In the wake of these decisions, many employers, rather than conduct validation studies, dropped selection testing altogether, while others cut back on test usage.
Which approach to validation is most preferred by Human Resource practitioners? Why? What is predictive validity and why is it preferred to concurrent validity?
Predictive validity within criterion-related validity because it helps rule out changes that may have occurred within the work group before study occurred (ex. poor performers already fired).
The characteristics of properly designed tests include reliability, validity, objectivity, standardization, and normative data. What are the differences between these concepts? In class we'll look at the bathroom scale used to weigh oneself. If you are buying a bathroom scale, which concepts would be most important and why?
Reliability
What is validity?
it actually measures what it purports to measure. If a selection test does not actually measure ability to perform a job, it is useless as a selection tool and, as stated previously, is vulnerable to successful legal challenge.
How effective are personality tests? Why?
often low in both reliability and validity. many require a subjective interpretation There are questions about the job relatedness of personality tests. Should be avoided.
How do the uniform guidelines define a "test"?
the Uniform Guidelines make abundantly clear, any selection device, method, or procedure, and not just psychological assessment instruments, is considered a test subject to the same validation requirements as paper-and-pencil tests
What is objectivity?
the scoring of test results. If all persons scoring a given test obtain the same results, the test is said to be objective. The most objective of all employment tests are those that use multiple-choice or true-false Subjective tests: Rorschach- applicant is shown cards containing inkblots and asked to offer an explanation of what he or she perceives in the inkblot. TAT- applicants are shown pictures of real-life situations and asked to provide their interpretations of each picture
What is standardization?
the uniformity of the procedures and conditions involved in administering and scoring tests. For the results to be useful, the test should be given to all individuals under conditions that are as similar as possible. Even the physical environment of the testing room should be the same for all test takers. Same materials, time limits, information, etc. should be provided.
What are some of the research findings about the use of employment tests?
(1) tests are more widely used in the public sector than they are in the private sector, (2) large to medium size companies are more likely to use tests than small companies, (3) larger enterprises are more likely to have trained specialists running their testing programs than are other companies, and (4) tests are more apt to be used to fill office and administrative positions than other types of jobs
What is the mental measurements handbook? Why is it important?
In searching for appropriate selection tests, one of the most important sources of information is the series of Mental Measurements Yearbooks. Almost all of the commercially available psychological, educational, and vocational tests printed in English are included in this series. Information on publisher, price, usage, as well as critical reviews by test specialists are included for each test. Any tests not discussed in the Yearbooks should be approached with a good deal of skepticism.
Why are cutoff scores subject to change and interpretation?
It is dependantupon the organization's selection ratio. During periods when there are a great number of applicants applying for jobs, higher cutoff scores will be used because the firm can afford to be more selective in the employees it hires. When there are fewer applicants, the cutoff score may be lowered so that job vacancies can be filled. Regardless of where they are set, cutoff scores should always reflect a reasonable expectation of success on the job.
What conditions must be present if a company would like to conduct a successful criterion-related validity study?
Technical feasibility Criterion Measures Representativeness of the Sample Statistical Relationships Operational Use of Selection Procedures Overstatement of Validity Findings Fairness
What do cognitive aptitude tests measure? For each cognitive aptitude, is there a profession that is more likely to use that aptitude? If so, what is it?
a person's capacity to learn or to perform a job that has been previously learned. Tests that measure this characteristic are most often used in the selection of employees who have had little or no job experience Verbal Ability- Managerial, technical, and sales positions Numerical Ability-essential in engineering, accounting, and similar jobs. Perceptual Speed-most likely to be used to ascertain clerical aptitude. Spatial Ability-Jobs that may require this aptitude include design engineer, tool and die maker, aviation mechanic, and assembler. Reasoning Ability-critical for executive, managerial, or sales jobs. General Intelligence-highly questionable today .
What is reliability?
extent to which it provides consistent results. If a person takes the same test several times, his or her actual scores will, of course, vary; this variance is referred to as the standard error, and is a measurement of the extent to which differences in scores are due to chance error. The closer the scores are to each other in repeated takings of the test, the more reliability the test has. Forms: test-retest, equivalent forms, split halves (best but difficult to ensure that the two halves are equal in content, difficulty, and nature)
What is normative data?
provides a frame of reference for comparing an applicant's performance on a test with that of a representative group of similar individuals who have previously taken the test. The test performance of the representative or sample group thus becomes the standard or norm by which the scores of future test takers will be evaluated or interpreted Typically, the scores from the sample group will be distributed according to a normal probability (bell-shaped) curve with approximately 68.3 percent of the scores falling within one standard deviation of the mean. Applicants scoring within one standard deviation are considered average relative to the sample population. Applicants scoring outside the range of two standard deviations would be considered to be unlikely performers (especially good or bad)
