2016 was not a good year to be a celebrity. At points it felt like every day we were losing a beloved famous name, from David Bowie on January 10th all the way through to Carrie Fisher and her mother at the end of December. “Damn you 2016!” became the cry of despair from all of us, as it felt like too many of our heroes were leaving us.
But was 2016 really as bad as it seemed? Back in May I analysed the first quarter of the year to see if the number of deaths was statistically unusual. And now 2016 is over I’ve done the same thing for the whole year. So with some confidence I can claim that yes, 2016 WAS a really, really bad year for celebrity deaths.
Crunching The Numbers
So here’s a recap on how I carried out this study. First, I needed a dataset of famous people who have died. After all - how can we judge who qualifies as famous? And how could I find a dataset that feels relatively unbiased? I could have, for example, taken data from IMDB - but then that would be biased towards Hollywood actors, and wouldn’t account for famous sports people.
So in the end I decided to go with the names listed on Wikipedia’s death pages. The thinking here is that in order to get a Wikipedia page, you have to meet certain notability guidelines. This is obviously relatively subjective, but the bar is sufficiently low that it’ll list everyone from Bowie through to minor British stars - and each inclusion will have been scrutinised by the wisdom of Wikipedia’s crowds.
But then there’s also another problem: How to gauge how famous someone is. Prince might have sold millions of albums - but how can we compare his level of fame to Carrie Fisher, who starred in one of a number of the biggest films of all time? And how can we compare both to Fidel Castro? That is surely the definition of comparing apples to oranges.
And this is where I hit upon pinching an idea from Google. One of the ways Google decides the order in which to list search results is by looking at the number of pages linking into it. The more links to it, the better the source is thought to be. So I decided to analyse the number of other Wikipedia pages linking in, to use as a proxy to measure the level of fame. For example, 2209 other Wikipedia pages link to Muhammad Ali’s page, but only 170 other pages link to that of Liz Smith, the former Royle Family star - implying that the boxer is many times more well known than the actress.
As I noted last time around, obviously Wikipedia isn’t perfect - it has its own biases and different topics could be covered unevenly. For example, the English wiki is mostly biased towards the English speaking world. But I’d still argue that the data is still good enough for analysis - and anyway, the English bias specifically probably even better reflects our own biases about who counts as a celebrity.
Overall, I’ve processed the data on the 39,313 people who Wikipedia lists of having died over the six years 2010-2016. So, caveats and methodology out of the way, let’s crunch some data and see just how bad 2016 really was.
What the Data says
So, first of all, here’s a chart showing the number of individual celebrity deaths across each year. The X-axis shows the number of days into the year, as because of leap years, dates aren’t directly comparable.
As you can see, 2016 was by a small margin the worst year for celebrity deaths - followed by 2014, then 2015. Earlier years, by contrast, were somewhat less deadly.
Now what happens if you factor in the notability scores I mentioned - the number of other wiki pages linking in, to give an indication of just how famous each individual was. Here’s all of the deaths plotted across the year, with each day’s total “notoriety” score plotted on the Y-axis.
The years are plotted on top of each other - and it’s interesting to see the spikes. As you can see, on January 10th 2016, there is a massive spike - of 5870 notoriety points. David Bowie accounts for 5546 of these points, and the other celebrities that died on the same day only had 324 points between them (Bowie is actually the most famous individual to have died across the entire six year period, according to my method). Some of the big spikes - such as the two 2015 spikes in the middle, cannot be attributed to any one individual, but are the result of multiple moderately famous people dying on the same day.
As with the chart I created in the piece last May, this also shows one of the potential biases of using Wikipedia: The massive score of film critic Roger Ebert in 2013. According to this model he was nearly as famous as Margaret Thatcher (who died just days earlier the same year), but in reality this score is due to thousands of film pages on Wikipedia citing Ebert’s review in the “Critical Reception” section.
Plotting these scores cumulatively though is where you can really see just why 2016 felt so shocking. On the Y-axis is the cumulative notability score for all celebrities from that year.
Here there’s an enormous gulf between 2016 and previous years. The collective fame of everyone who died last year is collectively significantly more than previous years.
The Big Names
One other thing I did last time around was try to separate out the really famous people, to see if anything distinct is going on there too. I did this by filtering to only celebrities who scored over 500 notability points. This means that it tended to cut out obscure footballers, and minor actors and politicians and so on. Even Ronnie Corbett didn’t make the cut - because he’s only famous in Britain.
Here, something interesting happens when you look at the number of individuals who died (not their cumulative scores):
That’s right - of the bigger celebrities, both 2014 and (narrowly) 2015 was a more deadly years. These are the likes of Leonard Nimoy, Terry Pratchett and Cilla Black.
Though even more interestingly, when you then use the scores of these 500+ scoring people, 2016 grabs an enormous lead once again.
This suggests that not only did we lose the most celebrities of recent years but many those who died were significantly more famous than the people who have died on previous years.
And that’s probably why 2016 really did feel like things were so bad. It wasn’t just Brexit and Trump that put the year on a downer: Yes, the people we love (and, er, Fidel Castro) really were dying faster than ever - and those that did die were some truly massive stars, which made it even more dramatic.
Here’s hoping that the Grim Reaper will be slightly more forgiving in 2017.