Assuming Tools For Digital Magazines

In  Daniel H. Simon and Vrinda Kadiyali tell us about their methods:

We consider three alternative approaches for handling the missing information on digital content for the pre-1996 period. First, rather than dropping the 97 observations for those titles most likely to have had a website prior to 1996, we impute their digital content for earlier years using a simple rule: we assume that for any years between 1993 and 1995 in which a magazine, which offered digital content in 1996, had registered a domain name, it offered the same level of
digital content that it offered in 1996. In other words, we simply fill in the 1996 level of digital content for each of the years between 1993 and 1995 in which a magazine had registered a domain name.  

Uh, dudes, do you remember the tools we had in 1993?  Neither do I. Bartender, more Wine!.  There is no way a magazine could have maintained a website in 1993 like it did in 1996+.  If that makes sense.  Which doesn't mean their data/conclusions are wrong, just that this is a bad assumption.