|•||Consequences of Unreliability|
This chapter extends the discussion of measurement reliability introduced in Part II. Two points introduced in that part are summarized here. First, scores from measures of constructs invariably contain random (unsystematic) error. Reliability, the opposite of random error, refers to the systematic or consistent portion of scores. Reliable measurement is thus necessary for construct valid measurement.
Second, there are three contexts in which reliability is often of interest to researchers. These three differ regarding the type of reliability addressed. Internal consistency applies to multiple items in a measure, interrater reliability applies to multiple observers or raters, and stability applies to multiple time periods. In each case, reliability means that scores obtained across