Are you sure that the passwords that were the object of the study came from the password manager itself? The paper that is referenced says this in the introduction:
> In this paper, we seek to fill in the gaps by gathering and analyzing a large collection of leaked password datasets across multiple years and various online services
> Virginia Tech researchers led by Dr. Wang have collected a number of publicly available password datasets from the Internet in January 2017. The datasets were obtained from various online forums and data archives.
It looks like they just used already leaked passwords.
