ABSTRACT
Basically, item analysis is concerned with the problem of selecting items for a test so that the resulting test will have certain specified characteristics. For example, we may wish to construct a test that is easy or one that is difficult. In either case it is desirable to develop a test that will correlate as high as possible with certain specified criteria and will have a satisfactory reliability. The index of skewness should be positive, negative, or zero for a specified population. If a battery of several tests is being constructed, it may be desirable to have the intercorrelations as low as possible. It is also of considerable interest to be able to construct a test so that the error of measurement is a minimum for a specified ability ränge or so that the error of measurement is constant over a wide ability ränge, as is assumed in the development of formulas for Variation in reliability with Variation in heterogeneity of the population (see Chapters 10, 11, and 12). In each of these situations it would be convenient to be able to write the prescription for item selection so that we should be able to subject a set of K items to an appropriate type of analysis, and then to select the subset of k items that would come nearest to satisfying the desired characteristics.
