The overall concept for chapter six concerns estimates of reliability for tests. This includes a review of measurement scales (nominal, ordinal, interval, and ratio), steps for the selection and creation of tests, and some of the more commonly used forms of reliability estimates (Cascio & Aguinis, 2011). Within the steps for selecting and creating tests, conducting an item analysis using item response theory (IRT) is also discussed. In addition, interpretation of reliability is included in the chapter, and the range of individual differences, some of the difficulties in measuring reliability, size and representativeness of a sample, and statistical measurement of standard error (Cascio & Aguinis, 2011). Scale coarseness is another important …show more content…
Validity and reliability are the foundation of evaluating tests, and both are needed for a test to be most effective. The three primary methods of validity are discussed in this chapter, and include content-related, criterion-related, and construct-related methods for gathering evidence of validity (Cascio & Aguinis, 2011). Each method is discussed in depth, for example, concurrent and predictive validity for criterion-related evidence. Content-related and criterion-related evidence of validity are the most commonly used methods for tests in work environments, with content-related edging out criterion-related if sample sizes are too small or if there are range restrictions (Cascio & Aguinis, 2011). Although, criterion-related lends statistical power to its evidence for validity, which is lacking in content-related evidence. Construct-related validity provides foundational evidence for the interpretation of test scores, making it necessary for other methods of gathering validity evidence. However, according to Stults (as cited by Cascio & Aguinis, 2011), construct-related evidence is limited by 1) a lack of quantifiable criteria, 2) an inability to consider differential reliability, and 3) an implicit assumption beneath the procedure. Methods for gathering evidence is also outlined and explained in the last section of the chapter. Sometimes, if local validity is not possible (e.g. too small of a sample), other methods such as synthetic, test transportability, or validity generalization (Cascio & Aguinis,