In this recent post  on how improvements in the iPod make price changes hard to measure accurately, I closed by wondering how the BLS handles such challenges in measuring prices and inflation. In the comments, Spencer writes:
You do not have to wonder.
Give them a call and they will be more then happy to explain their methodology.
When you know how they do it you can come back and explain what is wrong with what they do.
But just writing a post implying they ignore these issues is an extremely biased posting.
You might also check on the weight of iPods in the index. It is
probably something like 0.001% so even completely ignoring the quality
improvement in the iPod would not have a significant impact on the
reported inflation rate.
Actually, I was being rhetorical (or maybe the right phrase is tongue-in-cheek) in wondering about the BLS. I should have posted a link to this post  where I cite studies (bottom half of the post thought the first half is another important source of measurement error) of how the BLS tries to measure quality changes and the non-trivial magnitude of the inevitable errors that result.
This isn’t a slap at the BLS. They are diligent and I assume they do the best they can. But in a productive economy like ours, it’s too expensive to measure the impact of quality changes with any precision. They happen too quickly and too often. The important thing is to understand the limitations of our inflation measures and what they imply for the standard of living of the average American over time.