Well, it is really not clear what specific variables and codes they are using to create the figure. I have a couple guesses. First, I seem to get frequencies relatively close when I sum WKSTAT codes 21 and 22 together. These are both part-time for economic reasons categories, only they differentiate between “usually part-time” and “usually full-time”. Second, there is another variable (WHYPTLWK) that indicates reason for part-time status, but the categories are much more detailed. I doubt they are using this variable, but I wanted to let you know about it, just in case. Third, it seems they are seasonally adjusting their data, which is not done with IPUMS CPS data.
A final note is that we generally don’t expect to exactly replicate published statistics. We expect to get close, but due to restrictions placed on public use data, an exact replication is rare.