A More General Sufficient Condition for Partition Testing to be Better than Random Testing

Partition testing is an approach to program testing in which the input domain of the program is divided into partitions and test cases are selected from each partition. An alternative approach, known as random testing, is to select test cases at random from the entire input domain. Weyuker and Jeng (1991) observed that if all partitions are equal in sizes and the number of test cases selected from each partition is the same, then partition testing has a better chance of detecting at least one failure than random testing. This condition has been generalized by Chen and Yu (1994). They proved that partition testing is better than random testing so long as test cases are selected in proportion to the size of partitions. In this paper, we prove a more general sufficient condition. Partition testing performs better if the sampling rates are higher for partitions with higher failure rates. Some special cases that follow from this result are also considered.