Psychological Testing And Assessment We Learn
In the Text Of Psychological Testing And Assessment We Learned About
In the text of Psychological Testing and Assessment, we learned about three most common and different measures to determine the reliability of an assessment. They are the test-retest reliability, parallel and alternate forms of reliability, and the internal consistency reliability. All three measures have coefficients that are interpreted differently from one another.
The test-retest reliability coefficient is the average of two scores of the same test (Cohen, Swerdlik, & Sturman, 2018). A coefficient of =.50 is a very low reliability. This coefficient could mean that a score of 100 was graded and a score of 0 was graded on the second test, or vice versa, averaging a 50. A coefficient of =.50 means it is unacceptable because of the great variance between tests. The test-retest method would be useful in determining descriptive traits that are stable over time, such as personality or intelligence.
When considering the alternate forms of reliability, this involves using two different versions of a test (Cohen, Swerdlik, & Sturman, 2018). For the reliability to be high, the difficulty level must be equivalent across both versions. The parallel forms reliability coefficient of =.82 indicates that the reliability is moderate to high, functioning as a standard benchmark. This coefficient suggests that the two test forms are sufficiently consistent with each other, making them suitable for assessment purposes that require minimizing practice effects or cheating. However, a potential source of error variance in this measure is the difference in difficulty levels between the two test versions, which could affect reliability scores.
Internal consistency reliability is a measure that does not require repeated testing or multiple test versions (Cohen, Swerdlik, & Sturman, 2018). It assesses how closely related the items on a test are to one another, indicating the consistency of the instrument. This measure uses coefficients such as Cronbach's alpha, where higher values reflect greater internal consistency. For example, a coefficient of =.92 indicates very high reliability, suggesting that the items are highly related and measure the same underlying construct. Internal consistency is especially useful for tests like rating scales or questionnaires where uniformity of items is critical to validity.
The reliability coefficients discussed—test-retest, parallel forms, and internal consistency—serve as essential indicators of a test's dependability and validity. Each has unique applications and limitations, and the selection of an appropriate measure depends on the nature of the construct being assessed and the context in which the test is used. Proper interpretation of these coefficients ensures that assessments yield meaningful and consistent data, ultimately supporting accurate decision-making in psychological testing.
References
- Cohen, R. J., Swerdlik, M. E., & Sturman, E. D. (2018). Psychological testing and assessment: An introduction to tests and measurement (9th ed.). New York, NY: McGraw Hill.
- Carmines, E. G., & Zeller, R. A. (1979). Reliability and validity assessment. Sage Publications.
- Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). McGraw-Hill.
Principles and practice of structural equation modeling. Guilford Publications. International Journal of Medical Education, 2(1), 53–55. Journal of Psychological Methods, 19(3), 255–273. Measurement reliability in psychological assessment. Academic Press. Discovering statistics using SPSS (4th ed.). Sage Publications. Psychological Monographs: General and Applied, 68(1), 1–32. Multivariate Behavioral Research, 27(3), 377–394.