I can't help but wonder if knowing that they were being watched didn't affect their behavior.
For example, I'm a pretty competent programmer. I don't usually do very well on a whiteboard or someone else's computer during an interview. But day-to-day I turn out plenty of code.
To presume that observation has no effect on the outcome seems a little naive to me. Stranger still that an academic seemingly doesn't realize that.
Also consider that such people are typically integrated with the community they are studying for months, even close to a year. We are certainly different when observed, but we also get used to a new presence, eventually consider them "one of us", and will resume (mostly) normal behavior.
They're not technically academic works, but I consider David Simon's books "Homicide: A Year on the Killing Streets" and "The Corner: A Year in the Life of an Inner-City Neighborhood" (co-written with Edward Burns) to be in the category of such studies. The first book was the inspiration of the television series of the same name, and both were inspiration for The Wire.
"The Corner", by the way, is one of the most emotionally devastating books I've read, and I tend to pick very unhappy topics.
I have no issues with ethnography being used to broaden our understanding, but the understanding should be there that this is not science (and should not be called science) and the biases of the author have more of a chance to come into play.