Why commuters do not read

One of the curses of my new job is having to commute from Cambridge into London two or three (or four or five ...) days a week. Commuting must be good for something. One of the things I find it good for is primate behaviour research. I have found, for example, than commuters do not read books.

This started with an observation last November - lots of people on the tube were starting books. Lots of them, reading the first few pages of books. None of them reading the end. Surely just coincidence?

Think again.

I started collecting statistics. I observed all the people on the trains that I saw reading books, and wrote down how far someone was through a book. I could not tell whether they were on page 276 out of 327, but I could estimate what proportion of the book they had read - 30%, 70% etc. Only real books count - manuals and computer books don't, as people do not read them linearly. Magazines etc. don't count, mainly because it is impossible to tell whether someone is on page 7 out of 13, or page 9. But a meaty bit of Tom Clancy or Dostoevsky or molecular biology or something, I got quite good at estimating how far on the readers had got. Of course, I had to note all the books being read in a carriage, to get a valid sample. This lead to much craning and staring, and in any other country in the world I would probably have been shot. In England, of course, no-one comments.

Anyway, here are some numbers, collected into suitable bins. I note what month I made the observation in, and a summary of the fraction of the book read. These bin sizes are selected so that my collection categories (which tend to run in decades) do not bias the results too much.

% read Nov. / Dec. Jan. / Feb. Mar. / Apr. May / Jun.
0 - 33 12 27 25 12
33 - 66 8 13 12 6
66 - 100 0 13 9 8
total 20 53 46 26

(The 'total' statistic is not significant, as it reflects only how many times I travelled into London and how many times I had a piece of paper to hand to write numbers on, and whether I went to sleep instead.)

And, sure enough, about half the people on the train are starting books, and hardly any are finishing them. Chi squared result on testing that people should be randomly reading the first, second or third third of their books = 30.185 (9 degrees of freedom, p<0.005). There is no real bias between the train going into London in the morning and going out in the evening (Chi squared = 1.05, p>0.2), nor is there any apparent bias towards more front-end-book readers in January as opposed to December. This skuppers the only reasonable theory, that people start reading books in the morning and have not finished them by the evening, or that people start reading books after Christmas and have not finished them by the summer when I write this. This leaves the only possible explanation that people actually only read the first 1/3 of books on average.

The implications of this for the authors among you need not be spelled out. For the publishers, think on this. You can build a short story into a block-buster saga simply by adding 200 blank pages on the end. Half your readers will not notice, and you can simply give the other half their money back, claiming a printing error.

I look forward to seeing this in W.H.Smiths.

June 1997