That research is incomplete. Look at the bottom of page 3:
"A portion of the population without a drivers license – whether valid or not – will have a photo ID,
but without an analysis by race and location, it is not possible to estimate that population. "
The question here isn't whether or not they have valid drivers license, it's whether or not they have a valid photo ID good enough to vote with.
Thank you for providing that link, though. It's less biased than the others so far.