Adaptive Random Testing (ART) is an enhancement of Random Testing (RT). It is known when ART can outperform RT and when it cannot. Previous studies assumed that the test cases are selected with replacement. It is unknown whether selection without replacement can enhance the effectiveness of ART and also RT. Our studies include whether ART outperforms RT because it evenly spreads test cases or it generates fewer duplicate test cases, whether the input domain type and size have an impact on the effectiveness of ART, whether ART can still outperform RT when there exists only one single failure-causing input (in that case, failure-causing inputs do not cluster together). This study comprehends our understandings about whether ART is really better than RT.