Ben Lorica, over on O'Reilly Radar, published an interesting piece this morning entitled "Long Tail iTunes Book Apps Are More Expensive." His analysis shows that of "book apps" in the app store, those in the long tail (less popular ones) tend to be more expensive. He demonstrates this by showing that as you go further and further into the tail, the cumulative mean of the app prices goes up.
I thought I'd try to recreate the effect using other products, and something less high-tech, to see if the effect holds. I decided to look on Amazon for the price of books in the US History section. Amazon has these sorted by popularity, which makes it easy to run the calculation. Check out the results for prices (sorted from most popular (left) to least) and the cumulative mean, median, and standard deviation:
What can we learn from this? Well, for starters, the same effect (thought very slight) shows up here. In the second graph, the blue squares are the cumulative mean, and you can see that after they settle out (some initial noise due to low sample count), they do in fact rise slightly with decreasing popularity. Interesting.
However, it is also clear that the cumulative median is almost flat. That means (roughly) that your "average book" is still priced the same, but there are more high outliers. This also shows up in the rising standard deviation.
While this trend appeared interesting at first glance, I have two criticisms. First, shouldn't we expect that if it's difficult to know quality a priori, cheaper will be more popular? Second, at least in the case of books, the trend appears so weak as to be almost irrelevant.