Essentials of Psychological Testing by Urbina Susana

Author:Urbina, Susana
Language: eng
Format: epub
ISBN: 9781118680483
Publisher: Wiley
Published: 2014-08-04T00:00:00+00:00

Evidence Based on Relations to Other Variables

Convergence and Differentiation

Correlations Between Different Measures. A simple and frequently used way of gathering evidence that a particular test measures the construct it purports to measure is by establishing high correlations between its scores and those of other instruments that are meant to assess the same construct. One of the most basic examples of this type of procedure occurs when tests are revised and renormed. In such cases, test manuals almost invariably cite high correlations between the new and previous editions as evidence that both are measuring the same constructs. This is somewhat akin to computing correlations between alternate forms of a test to establish the reliability or consistency of scores across different forms of a test. It may be recalled from Chapter 3, however, that even if correlations between the old and the revised versions of tests are very high, the normative scores for restandardized versions tend to drift in one direction or another due to changes in the population at different time periods.

In a similar fashion, test developers typically present data that illustrate the relationship between the scores of their tests and those of comparable instruments as evidence of score validity. For instance, the manuals of most major ability scales cite correlations between their scores and those of the other well-established instruments of the same type. By examining these data one can learn—for example—that the correlation between the WISC-IV Full Scale IQ and the Mental Processing and Fluid-Crystalized composite scores of the Kaufman Assessment Battery for Children, Second Edition (KABC-II), for a sample of 30 Native American children tested with both instruments, were .86 and .84, respectively (A. S. Kaufman & Kaufman, 2004). Correlation coefficients of this magnitude are typical for ability scales that assess comparable constructs and serve to corroborate the fact that a good deal of the variance in the scores on these tests is shared.

Correlation coefficients can also be obtained across the scores of subtests from different scales. A typical example of this in the area of abilities would be the correlation of .71 obtained between the Delayed Memory Index scores of the Wechsler Memory Scale-Fourth Edition (WMS-IV) and of the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) for a sample of 100 examinees (Wechsler, 2009). In the realm of personality measures we find many equivalent examples such as the correlation (r = .68) between the Depression scale scores of the MMPI-2 and those of the Dysthymia scale of the Millon Clinical Multiaxial Inventory–III (MCMI-III), or between the scores of the Beck Depression Inventory and those of the MCMI-III Major Depression scale (r = .71; Millon, Millon, & Davis, 1994). As might be expected, correlation coefficients calculated across various types of tests and subtests abound in test manuals and in the psychological literature, even though these indexes often are not very cogent or informative.

Intertest correlations are as ubiquitous as they are because the data for correlational studies on small convenience samples are easy to gather, especially for paper-and-pencil tests that can be administered to groups readily.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.