ABSTRACT
Replication is important because a proposition which is confirmed at one time and in one place will not necessarily hold at other times and places. Testing and retesting a proposition at different times and in different places provides more conclusive results. The selection of departments for more intensive study was deliberately designed to exclude those that lacked an adequate population base to provide a significant amount of use data, had a large proportion of part-time students, whose borrowing patterns were likely to be inconsistent with those of full-time students. The principal factor that determines the reliability of the negative binomial distribution (NBD) model as a predictive tool appears to be the nature of the use trends at the point in time that is selected as the base period for fitting the observational data to the model. Circulation data were accumulated weekly and input into the NBD model.
