To select a sample size she selects a confidence of 92% relying on the feeling that one chance in twelve presents an appropriate risk of being wrong. The collection has never been weeded so she believes that ten or fifteen percent of the books might be unused. Since the most conservative choice is to assume a high value for variability, she chooses 15%. She thinks that a relatively small precision interval will be important in convincing administrators to support a weeding program, so she sets a value of 3%. The sample size is 433.
Because all catalogue cards are a standard thickness the sample can be selected by measurement.
Category 1 Drawer numbered from 1 to 38
Category 2 Inches numbered from 0 to 14
Category 3 16ths numbered from 0 to 15
Not all drawers are full of cards so she saves the raw file so that she can draw one or more supplemental samples.
To actually draw the sample a clerk pulls the specific shelf list drawer and holds the cards forward to make the drawer relatively snug. She then lays a ruler on top of the cards and gently inserts a blank card at the measurement given in the sample table. If the card is a book card, author and call number information are recorded on the sampling sheet if the card is for a serial or series or serial or if there is no card at the measurement the sample item is marked for deletion. Ordering the sample by drawer and distance makes the task of collecting this information go quickly.
this page is at http://testbed.cis.drexel.edu/sample/example3.html