ABSTRACT
This chapter addresses some commonly encountered misconceptions and misuses of psychometric techniques. It also addresses issues such as whether scores on tests which comprise several sub-scales and which have coefficients alpha in the order of 0.95 are really as accurate as many practitioners assume. The chapter explains the issues arose during conversations with my psychometrically inclined ex-colleagues in educational and clinical psychology, who focus more on the interpretation of individual scores than on the correlational analyses typically used by researchers in individual differences. The problems with using coefficient alpha as a measure of reliability should be carefully considered before using it to estimate confidence intervals. Before considering how to calculate change scores from individuals, it is important to outline one design which occasionally creeps into journals. Increasing the standard deviation of the error term is equivalent to reducing the reliability of the scale. Knowing a test's overall reliability is enormously important for practitioners.
