Judith Chevalier and Dina Mayzlin have studied the impact of consumer reviews of books on word of mouth and subsequent sales, here is their NBER working paper, here is another draft. They find the following:
1. Most consumer reviews of books on Amazon.com and Barnesandnoble.com are very positive.
2. The reviews at Amazon are longer and more extensive. They are also more critical on average.
3. Better reviews on one site boost relative sales. The use of two sites gives us a controlled experiment to determine that word of mouth does indeed help authors rather than being a mere side effect of higher sales.
4. A bad review hurts you more than a good review helps you.
5. It remains to be seen whether allowing consumer reviews increases aggregate sales or simply shifts around sales to more suitable titles. Even a shifting affect, however, may increase consumer loyalty to the on-line site. If you know that Amazon helps you discover good books, you may be more likely to buy from Amazon.
My advice: I don’t put much stock in how favorable the Amazon reviews are, whether I am buying books, movies, or music. (I am most likely to buy music from Amazon.) This well-known example is one reason to distrust the reviews, although I think bad taste is more common than masquerades. Instead I look at how many reviews have been generated. I take this as a kind of sufficient statistic for how much passion the item has generated. Since I am at the tails of just about any distribution of taste, and since most cultural products disappoint in any case, look for something that creates a spark in people. I then see some chance of finding a product that I truly love. This advice will sometimes steer you wrong, but a little added intelligence will allow you to make the necessary adjustments.
Thanks to Eric Crampton for the pointer to the article.