Testing software product lines (SPLs) is difficult due to a huge number of possible products to be tested. Recently, there has been a growing interest in similarity-based testing of SPLs, where similarity is used as a surrogate metric for the t-wise coverage. In this context, one of the primary goals is to sample, by using search-based algorithms, a subset of test cases (i.e., products) as dissimilar as possible, thus potentially making more t-wise combinations covered. Prior works have shown, by means of empirical studies, the great potential of current similarity-based testing approaches. However, the rationale of this testing technique deserves a more rigorous exploration. To this end, we perform a correlation analysis to investigate the internal relationship between similarity metrics and the t-wise coverage. We find that similarity metrics generally have a significantly positive correlation with the t-wise coverage. This well explains why similarity-based testing works, as the improvement on similarity metrics will potentially increase the t-wise coverage. Moreover, we explore, for the first time, the use of the novelty search (NS) algorithm for similarity-based SPL testing. The algorithm rewards "novel" individuals, i.e., those being different from individuals discovered previously, and this perfectly matches the goal of similarity-based SPL testing. We demonstrate that the novelty score in NS has a (much) stronger positive correlation with the t-wise coverage than the widely used fitness function employed in the genetic algorithm (GA). Empirical results on 31 software product lines, either realistic or artificial, validate the superiority of NS over GA, as well as other state-of-the-art approaches. Finally, we investigate how the performance of NS is affected by the ways of generating new products, and by its key parameters. In summary, looking for novelty provides an alternative way of generating test cases for SPL testing.