If the entire average viewership of CSpan and @50% of the total PBS viewership actually decided to forego their normal habits and watch major news events in something other than a major network, Trump would have tied Romney.
This comparison takes two equal sets of standards and compares them. To assume that had they included those missing sources, the results would have been different is highly illogical, since it presumes that the percentage of viewership watching those networks this cycle was significantly different than the last.
Remember, if you want to include PBS and CSpan this year, you'd have to include their numbers in the last cycle as well, and to argue that every sngle other outlet had a significant drop in viewership while those two had an equally significant increase, without any data to back that argument up, makes the entire premise that had they been counted the results would have been different totally invalid.