Reliability and Validity
Reliability and
validity are inter-related concepts central to the evaluation of any questionnaire used for measurement purposes. Simply put, reliability has to do with how consistently a test measures something and validity is concerned with whether a test measures what it purports to measure. Validity is dependent on reliability, in that you can not have high validity without high reliability. However, you can high reliability and poor validity if the test is measuring something consistently, but not what you think it is supposed to measure.
For purposes of this discussion, a
standized test consists of a collection of individual test items with some form of force choice response options. Questionnaires used in behavioral health outcomes research typically employ some form of a
Likert scale.
While there are a number of methods used to evaluate reliability and validity, the results are typically expressed in the form of a
coefficent of correlation ranging from 0 to 1.
Estimates of reliability are based on methods that attempt to correlate the test with itself, under the premise that if the test doesn't correlate with itself, it can't measure anything consistently. For more on estimating reliability, see
CalculatingReliability.
Estimates of validity are based on correlating the test with some other criterion measure or variable. For more on estimating reliability, see
CalculatingValidity.
Primary topics:
You may also view all topics in alphabetical order in
WebTopicList.
--
JebBrown - 06 Jan 2007