ABSTRACT
Score reliability characterizes the degree to which scores evaluate 'something' as opposed to 'nothing'. The English psychologist Charles Spearman (1904) first articulated a coherent measurement theory, sometimes called 'true score theory', but now often called 'classical theory', merely because this theory is older than the 'modern' measurement theories. The internal consistency reliability coefficients are computed by taking into account these item score correlations. A problem with split-half reliability estimates is that for a measure consisting of relatively few items there are a large number of potential item splits, each of which may yield different and potentially contradictory reliability estimates. The previously presented formula for Cronbach's α does not make clear the pivotal role that the correlations of the item scores have on internal consistency reliability. Poor score reliability compromises substantive research analyses by attenuating both pCALCULATED values and effect sizes.
