Here's the full list:
Back when I first came up with A18-49+ and the whole "league average" idea, the only years I had access to were 2006-07 through 2011-12. One thing observed in those years was that the league average decline seems to correlate pretty well with the strength of that season's new series. The general theory was that there was a certain "true league average decline." Within any individual season, a good new class could make that up to a couple points better, while a weak one could make it a couple points worse. For many years, this theory seemed to work if you assumed the "true" decline was something like 8-9%. It did a couple points worse than that with the awful classes of 2010-11 and 2012-13, and a couple points better with the great classes of 2009-10 and 2011-12.
The first real exception to this was in 2013-14, when the newbies were fairly strong but the league average decline was on the steep end of the range, almost exactly matching the previous season. I seem to recall mentioning this in passing, but perhaps it deserved more scrutiny at the time. And 2014-15, with an even better class (including Empire) saw yet another decline in that same upper-10 percent range.
Now that we've finally had a bad class again in 2015-16, it seems clearer that this theory about a "true" league average decline could still apply, but it has sped up. In this new world, maybe those results from the previous two years were actually good declines rather than bad ones. The silver lining for a rough season like 2015-16 is that it didn't just suddenly happen this year. The theory is this is actually the third year of a new era, where the "true" decline is somewhere in the vicinity of -12%ish.
To put some labels to it, you could say 2006ish to 2013 is the "rise of DVR/cable" era, with a "true league average decline" somewhere in the vicinity of -8.5%. A new era began in 2013-14, with DVR/cable still a factor plus the "rise of streaming/cord-cutting" tacked on. The "true decline" in this period seems to be somewhere in the -12% area.
It's also very possible that the steepening of the "true" decline is more gradual than I am trying to make it. After all, you can divide these declines into "good" and "bad" categories and see them getting a tiny bit steeper even within these "eras." Specifically, good goes from -6.4% in 2009-10 to -7.1% in 2011-12. Bad goes from -9.1% in 2008-09 to -9.7% in 2010-11 to -10.9% in 2012-13.
By the way, it's harder to know what to make of the years we have prior to the DVR's explosion circa the mid-aughts. It's pretty clear that the true decline is a lot slower; getting from the 2001-02 number to the 2005-06 one would take about a 1.5% decline on average. But on a year-to-year basis, the trends vary a lot more wildly, including a couple years of league average growth. Maybe some day we'll have more to look at!
Here's the now updated A18-49+ League Averages post.