Given the context the the article provides, and that the raw table does not, I am inclined to believe the pew research article, and would suggest the linked table needs to also have some methodology and definition of what it thinks it is and is not capturing.

The pew article also looks at statistics like "In 2019, 20% of households are shared households, up from 17% in 2007."

The weakness in the pew article is it's forecasting growth in household size from one year of its model, obviously we need to wait more time to see if it's model is correct.