Abstract
Implementing graphical and analytical procedures for developing parallel tests. We present an instrumental study in which two general procedures (graphical and objective) for constructing parallel forms are implemented. Both procedures are based on an item pool which is calibrated by means of the classical test theory. The graphic procedure is Gulliksen's matched random subtests method. The objective procedure is based on the criteria proposed by van der Linden and Boekkooi-Timminga, and uses zero-one programming. The stand-alone program FOR-PAR is free and user-friendly, and allows the parallel forms to be obtained directly from the item scores. FOR-PAR covers an important need in applied research, because developing parallel forms is a requirement in some applications, and, so far, no programmes of this type were available. The procedures which are implemented were applied in an illustrative example based on an adjustment test intended for visually handicapped people.