Given the significance of software testing, many methods such as black box testing have been introduced in order to make testing as efficient as possible. Experimental evidence is required to attest to the validity of such methods. In this thesis, we conducted experiments on twenty four test suites generated from two case studies targeting a real industrial system for generating financial reports. The experiments aim to evaluate the effectiveness of different Category Partition (CP) specifications on the same problem as well as that of the 'Single' and 'Error' constraints on the same CP specifications. The experiments also evaluate the effectiveness of the three selection criteria of Base-Choice, Each-Choice, and Pair-Wise. The effectiveness is measured in terms of cost, which is the number of generated test cases, fault detection of errors created by mutating code under test, and code coverage using Visual Studio.