Predictably Irrational

I recently read "Predictably Irrational", a 2008 book by Dan Ariely, then a professor of behavioral economics at MIT.  Based largely on his own experimental work it discusses situations in which people often behave in ways which are at least arguably irrational.  Unfortunately I didn't think the book was very good.

My main problem with the book is that it is weak on the big picture.  By way of analogy human vision is generally pretty good but not perfect as the existence of optical illusions shows.  But a book just describing various optical illusions would not be a particularly good way of giving an overview of human vision.  Similarly it is hard to know what to conclude from a book describing a few experimental situations (often quite artificial) in which people behave irrationally.  I didn't get much more from this book than the observation that people sometimes behave irrationally which I already knew.   

One reason for the book's problems with the big picture is as mentioned above it is largely based on Ariely's own experimental work.  I believe a book like this should present a general overview of the field suitable for the lay reader.  This would include summarizing the most important and well established experimental results.  Just describing your own work (in what I sometimes thought was excessive detail) is not as useful.  Among other things it is difficult to be objective about your work, its importance and weaknesses.  Also I suspect there is a publication bias at work in this whole field.  I doubt it is as easy to publish experimental results concerning situations in which people do behave more or less rationally.  Which could give an unbalanced view of how pervasive irrational behavior actually is.

The book does present an useful general principle, namely that people like to evaluate things relative to other things rather than on an absolute basis.  So  B may appear more attractive when presented with a clearly inferior alternative C than when considered in isolation.  So experiments can be devised in which people prefer A to B when given 2 choices but prefer B when given 3 choices with C an inferior version of B added.  The addition of C makes B appear more attractive although this violates models which assume A and B have a definite absolute value.  Similarly people tend to evaluate their circumstances relative to their recent past (or compared to people they consider their peers).  So happiness is not as related to income as much as you might expect.  A rich person whose life is getting worse will tend be unhappy while a poor person whose life is getting better will tend to be happy even though objectively the rich person is still much better off.       

There are a few more worthwhile observations in the book but not in my view enough considering its length.  And I didn't think the book was particularly well written.  So while I don't think the book is totally worthless I would not recommend it.