The main threat for our experimental design is that we
experiment only with two subject AUTs. The results may
vary for AUTs that have different logic or different source
code structures. This threat makes it difficult for us to generalize
the result that we obtained through experimentation.
However, since both applications are highly representative of
enterprise-level applications that they come from different
domains, we suggest that our result is generalizable for a
larger population of applications.