The use of test decks, to test the adequacy of programmed (and perhaps of built-in) checks, is not then primarily the task of the auditor. The programmer in fact can only state that he has finished the preparation of a program, after a test deck has run satisfactorily. It is the task of the auditor to make sure that the test decks used cover most of the situations which are to be tested.
21 Whatever is done by the auditor in the way of checking or auditing the program, he is not quite certain that in fact only the program which he certified is used, neither is he quite sure that the operator has not suppressed or added some instructions or biased some of the input data, unless a safe guarded automatic logbook is kept by the computer and made available only to the auditor.
22 Checking the program and using the processed information for auditing will thus generally not give the auditor full confidence in the unbiased operation of the system. This, however, is not peculiar to computer systems. Quite a lot of auditing procedures in them- selves are not sufficient to give full confidence. What is generally done in auditing is to take several complementary approaches to the audit of certain areas, and derive confidence from the combination. The overall View of the consistency of the principal figures is one important aspect.
23 The auditor could easily use a computer program of his own to test for consistency over a longer period than is done internally. This, of course, does not eliminate the need for ascertaining that the figures reflect the real state of affairs, because, however consistent a system of figures may be in itself, that alone is no safeguard that this system reflects what has really happened. The audit procedures to ascertain this are much the same as are used with manual systems.