Reliability and Validity in Classical Test Theory

Abstract

In CTTClassical Test Theory (CTT), reliabilityReliabilityis defined as the proportion of true score varianceTrue score varianceto total varianceTotal variance. It is most often estimated using the coefficient $$ \alpha $$. This index assumes the instrumentInstrumentis unidimensionalUnidimensionaland is not a testTestof unidimensionality. Construct validation addresses the substantive dimension of the variable assessed.