January 8, 2009
Follow me on Twitter

More Reading Journals

David Golding writes about reading logs, including Andrew Bowie's Art and Kate Orman’s Books Read. Art Garfunkel’s goes back to 1968. Golding's 2008 list is great, too ( he really should have seen Karen MacDonald in Mother Courage at the American Repertory).

There’s been a good deal of chatter in the past year about books as status objects: is it permissible, for example, to have books in your living room that you haven’t read? To casually leave Ulysses lying about, or to put Krugman on your end table so your friends will know you’re the kind of person who reads Nobel-Prize economics, even if you don’t? But web journals seem refreshingly free of this controversy, and Golding is as happy to pan Henry V and Pepys was to deplore Midsummer Night’s Dream.

Also fascinating is Golding’s survey of five years of reading.

Liked Ok Didn’t Like
2004 28% 26% 26% 18%
2005 44% 27% 16% 11%
2006 31% 28% 22% 16%
2007 32% 40% 14% 12%
2008 31% 31% 33% 3%

Golding then asks, as one ought, whether the variations are statistically significant, or not. We know that the books are not themselves normally distributed: he's not reading books at random, and even if you did pick random books off the shelf, those books have already been selected by editors and publishers and librarians. And we know that Golding’s reactions aren't normally distributed, either: three of his four categories are favorable. Still, there should be some disciplined test that says, "you're right: 2007 is actually different". If you remember your statistics better than I do, Email me.

Update: The answer, of course, is chi-square, which doesn't assume that the data are normally distributed — only that the errors are. While the error curves here might not be strictly normal, they’re probably close enough for chi square to be a plausible estimate.