Well, John Denver was flying an experimental plane with a known, potentially fatal design flaw. That’s not a plane crash that can be compared to any of the others.
Beyond that, I’m not sure we can say that the frequency has declined. You list 9 celebrities who died in the 40 years between 1962 and 2002, less two every decade. Then you note that Horner died just last year. So in the 13 years between 2002 and 2015 you have one death.
There isn’t any statistically significant decline. You’ve dropped from a long term average of ~1.8/decade to a new average of ~0.8/decade based on 11 sample points. Nothing to see there. Random chance explains such a modest decline with such a small sample.
Another point to consider is you sampling method. Your list excludes really big name celebrities like Carol Lombard or Will Rogers who died in the 40s. I’m sure we could find others form the 20s through to the 60s as well. Once you include these, you see that it’s not unusual to have one celebrity per decade die in this manner. You just locked onto a period that was insignificantly higher.
The other point to consider is your definition of a celebrity. You include Randy Rhoades and JFK Jnr on your list, but Rhoades is scarcely a household name and John John is best known as being someone’s kid. Not exactly major league celebrities. I’m sure that if we consider a definition of being famous that includes being someone’s relative and being a guitarist in a rock group with a single top 10 hit and two top 40, we could find others, including Steve Davis, Jenni Rivera, Lech Kaczynski and half the Russian hockey league. As with the “celebrities die in threes” meme, this one has a lot of confirmation bias. We remember celebrities who were famous in our formative years as being “big names”, so we remember JFK Jr and Rhoades but we ignore the deaths of celebrities of greater stature from before or after that period.