@dipped:

The short answer is, "I don't know." The analysis was based on an anonymized dataset for which I've obtained IRB approval from my university, so I'm trying to follow proper research protocol. I've done Phish stats for several years now--mostly using data pulled from publicly available data, and I was pleased that this work allowed the site Admins to trust me with the raw data.

That said, if you have econometric questions, send them to me via PM. The variables are (anonymized) User ID, show date, rating, and time of rating.

@greenarrow74:
Oh, man, I'd love to control for attendance bias! That's on my list of of things to ask the coders about. Last fall I proposed a split sample test of the "herding effect" during the NYE run (herding will be defined in Post #3), but it was simply going to be too much, too soon.

@grizzwold:

You do You, man.

Keep the comments coming