The relationship between criterion, test items, and instruction is discussed to show the problems of the pre‐test and post‐test evaluation of criterion‐referenced items. Six coefficients for selecting criterion‐referenced items, two of them developed by the author, are compared on the basis of two sets of data. There are differences in the usability in small samples, in the computability, in having limits for the range, and in the interpretation. The use of the Cox‐Vargas and the two Herbig indices seem to have some advantages.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados