We’ve had a couple of digitised books that proved really popular with online audiences. Perhaps partly reflecting the interests of the global population, they’ve been about prostitutes and demons.
I’ve been especially interested in how people have interacted with these popular digitised books. Imagine how you’d pick up a book to look at in a library or bookshop. Would you start from page one, laboriously working through page by page, or would you flip through it, checking for interesting bits? Should we expect any different behaviour when people use a digital book?
We collect data on aggregate (nothing personal or trackable to our users) about what’s being asked of our digitised items in the viewer. With such a large number of views of these two popular books, I’ve got a big enough dataset to get an interesting idea of how readers might be using our digitised books.
Focusing on ‘Compendium rarissimum totius Artis Magicae sistematisatae per celeberrimos Artis hujus Magistros. Anno 1057. Noli me tangere’ (the 18th century one about demons) I’ve mapped the number of page views (horizontal axis) against page number (vertical axis, with front cover at the top), and added coloured bands to represent what’s on those pages.
Observation 1: People are super enthusiastic at the start of a book, and less so as they get to the end
We can see that the general trend is that page views decrease as page number increases. i.e. people looked at more pages at the start of a book than the end. The most popular page is the title page, which also contains the first illustration. The first section of images has been viewed more times than the second section of images later in the book.
This is a trend that continues across other books, and maybe resonates with the idea of flipping through a book by hand: you look at the first few pages in detail, and stop looking once you’ve decided whether it’s for you or not.
Observation 2: People like looking at pictures more than text
You can see very quickly that the sections of pictures (pink areas), have been viewed a lot more times than the pages of text (blue areas). The fact that the text is in Latin and German and our audience is mainly English speaking may have something to do with that.
You might also notice that within the pictures sections that the spikes are for every other page. There’s one popular page, then a less popular page. That’s because each page with an image is adjacent to a blank page in the book.
This kind of pattern indicates that a lot of people seem to be metaphorically flipping through the book from image to image, generally ignoring blank pages, and sections that are plain text.
If you’ve had a look at any of our digitised books online, you might have noticed (or used) the thumbnails tab. It’s really the only comparable way to ‘flipping’ through the book by hand, as it allows you to see all the pages at a glance before committing to a particular page. This is the most likely way that the viewers of this book have so succinctly picked out pictures of demons to look at.
Interestingly, hidden within a block of written pages is a surprise image. This
man wizard page springs out of the graph as a sudden spike in page views. It seems that readers spotted him and thought he was interesting enough to take a closer look.
To conclude, the results really seem to correlate with what we’d associate with normal book browsing behaviour.
Clearly, when given the tools to skim a book, in this case in the form of the thumbnail views, users choose to skim rather than browse page-by-page. This raises some more interesting questions about how we can improve their ability to do this; whether there are any other skim-friendly tools yet to be developed.
I’m hoping to do more page view distribution analysis soon, specifically with more academic texts. Watch this space.