I don't think he was using anything incorrectly but you do do bring up a good point about data use. (The rest of this statement is speaking in generalized terms, not directed towards anyone here).
Data use can be extremely fickle and sensitive, particularly when using reanalysis data. It's always important that
1) The user has a strong understanding of the strengths/weaknesses of the dataset and
2) If the user is presenting writing a paper, the user clearly state these strengths/weaknesses.
There was a paper I was reading late last fall which was looking at the different versions of the ERSST dataset and I think comparing v6 versus v5 and a few previous versions and certain areas of the globe where there were noticeable discrepancies. And these discrepancies also had an impact on OLR anomalies and so forth. So its always important to understand strengths, weaknesses, limitations, etc. when dealing with datasets. I've really learned this the hard way doing so much with datasets and reanalysis from years as far back as 1900