Is it the same 1000 families from 2009, the article infers that it's the same 1000 families, whereas in reality there will be some people who drop out and some who come in.
Just like backward testing in stock, if you're measuring against the top X companies in 2000, they are not the same top 100 companies in 2018.