No abstract

What statistical methods are being used to identify biased test items? What are the advantages and disadvantages of each method? What sample sizes are required to use each method? Which methods are most useful to the practitioner?

No abstract

No abstract

Not infrequently studies are published concerning the prediction of grades in a single college for a single year's entering students.However, it is rare to find studied together a large number of colleges over several years' time. Without mere general study than that of the single c~llege for a single year, one can say little about what to expect in a new situation in terms of level of prediction, fluctuations from year to year, the gain to be obtained from using as predictors test scores in addition to records of previous performance in high school, and the differences in predictability for males and females.The data to be reported here are for all publicly-supported higher education in the State of Georgia for the five years from 1958 through 1962. There are IQ different colleges, but, since not all have students of both sexes, for each year there are only 35 different sets of data. For all five years there are 175 sets of data involving 27,961 different students. Data of such magnitude and exhaustiveness should be sufficient to permit general conclusions for much of public higher education, even though there has been no attempt randomly to sample from the entire nation.