Validity and Reliability Brainstorm

Topic: Validity and reliability - issues and discussion Date: 30 November 2006 12:51 PM
Subject: Validity and Reliability Brainstorm Author: Mondy, Steven

I've read the Hughes book as well as a lot of both Bachman books and what seems to be the case is that they present a very good definition of both construct validity and reliability and some of the problems associated with them, and give some comparison in the summaries at the end of the respective chapters. Though, I think it may be up to us to come up with a good idea of how they inter-relate, based on what we have learned about the essence of each.
Personally, I've found the Bachman books (Fundamental considerations in testing and Language testing in practice) more helpful for understanding the concepts, as they go into greater detail (a great part of those books is in the material that we were supplied with).
Maybe if we can start a brainstorm on this discussion board of the major issues involved in looking at how they inter-relate, or affect each other. Let me start listing a few ideas that I have come across in the readings…(please correct me if I’m way off base)

1. Reliability and construct validity are concerned with (a) minimizing measurement error (b) maximizing the effects of the language abilities that we are trying to measure. (Bachman, Fundamental considerations in testing, p.161)
2. Reliability is a necessary condition of construct validity, and hence for usefulness. (Bachman, Palmer, Language testing in practice, p.23). Bachman states in the summary on page 289 (Bachman, Fundamental considerations in testing, p.171) that validity is one of the most important qualities to consider in how we use tests, but reliability is necessary to determine potential sources of measurement error and for being able to predict the effects on test scores.
3. Reliability can be defined in terms of CTS theory (classical true score measurement theory) in terms of the correlation of true scores on tests (Bachman, Fundamental considerations in testing, p.171). As long as there is consistency between the scores of various candidates, we can be more certain of item reliability, and therefore of reduced error in testing technique.
4. Please correct me if I’m wrong, but validity seems to be concerned with having a test that closely matches the circumstances of the candidates taking the test, and creating a testing environment that allows for the best possible results to be obtained.
5………
6………
Can anyone add to the list…
Steven Mondy