Kindle Sales, Print Sales, and Spurious Correlations
Remember a few weeks back when everyone echoed the NYTimes' mistaken report about the decline of ebook sales?
That story proved the validity of the saying that a good fiction can go halfway around the world before the truth has a chance to get its pants on, and we saw that proof again this week when news came out that Waterstones was removing Kindles from its stores.
First The Bookseller, and then The Guardian and a bunch of other sites, reported that Waterstones m.d. James Daunt was dissatisfied with Kindle sales: "Sales of Kindles continue to be pitiful so we are taking the display space back in more and more shops".
The Bookseller went on to link the decline in Kindle sales to the resurgence of print:
The Waterstones move to remove Kindle devices from stores comes after figures from Nielsen Bookscan show sales of print books for the first 36 weeks of 2015 were up by 4.6% (worth £739.5m) when compared to the same period in 2014, the first time the print market has seen year-on-year growth at this stage of the calendar year since 2007.
Booksellers have suggested a range of reasons for the recovery, naming the economic bounce, publishers “upping their game”, booksellers sharpening their offer and, chief among them, the belief that the e-book “threat” is “vanishing".
The details about print sales were repeated by many sites, including The Guardian, and few checked to see whether there was a connection.
If they had then they would have noticed that the Bookseller doesn’t actually explain the connection between Kindle sales and print sales. In the absence of that explanation what we have here is an example of a spurious correlation.
As explained by Wikipedia, "spurious correlation" is a term coined by Karl Pearson to describe the correlation between ratios of unrelated measurements that arise as a consequence of using ratios, rather than because of any actual correlations between the measurements (you can find examples on this site).
I am misusing the term slightly in this post; I would describe this as the post hoc variant of a spurious correlation. There’s no correlation mentioned, but the use of two unrelated details implies a connection which doesn’t exist.
And to be clear, there is no correlation between the decline of Kindle sales and the sales of print books. Ask anyone who is following industry trends, and they’ll tell you that while we don’t know how many Kindles have been sold, ereader sales peaked in 2011 or 2012 and have been declining ever since.
Print sales, on the other hand, haven’t been going through a three-year-long revival (the AAP stats from earlier this week proves that beyond a shadow of a doubt – for the US, anyway; the UK is a different market).
So why were Kindle sales down at Waterstones?
To be honest, we lack the information to accurately answer that question. But I do like how Mike Cane independently reached the same conclusion I did:
Why is this any surprise?!
People have large phones to read on.
And better tablets.
That’s more or less what I wrote when I reported on this story earlier this week, and I think it’s the most likely explanation.
But John Gruber of Daring Fireball suggested an equally valid alternative when he linked to the story this week:
Easiest explanation for this is that Kindle users are Amazon users, and Amazon users buy their Kindles direct from Amazon. I’ve owned a few Kindles over the years, and it never even occurred to me to buy one anywhere else.
And it is just as likely that the decline in Kindle sales at Waterstones are due to an entirely different cause.
Your guess is as good as mine.
image by kodomut