(eg, they include college age students when they are usually left out because things like increase college enrollment would skew the numbers).
All their distributional data and numbers for what counts as low wage workers comes from a 2012-2016 Census Bureau survey. They then claim without much justification "we think it is unlikely our finding would change significantly if we considered more recent wage data" for the 2017-2019 period.
edits:
strangely enough they exclude graduate and professional student, but still include 18-24 year olds in high school or college. And they exclude self-employed for reasons similar to why you would want to exclude 18-24 year olds. And that 18-24 year old group accounts for only 4% of mid and high wage work, but 24.3% of low wage work, so it is clearly skewing the results.
And the largest category of low-wage workers (a plurality) is classified as "low-wage workers in a family with mid- to high- wage workers". I'm not really sure we care about that as much.
Reading more, they have some very strange inclusion/excisions, such as they exclude professional and grade students because they are seen as having a career path, they include "springboard jobs" because some might not transition out of them. That's just sloppy and lazy.
Sounds like some serious data hacking going on.