[this post is contributed by Phish.net user Paul Jakus, @paulj, who continues to contribute to the site in different ways. Thank you Paul! – ed.]

It seems unbelievable, but in 2018 Phish will embark on their tenth year of touring since returning from their breakup. Following last year’s Baker’s Dozen run, many speculated that we may have reached the peak of Phish “3.0.”

Setting aside one rain-shortened concert (7/19/2013), there have been 358 shows since March 6, 2009. The average show rating on Phish.net was 3.944 (out of 5), but the lowest-rated event was Grand Prairie’s 2.377 (10/25/16). If Phish reached “bottom” as recently as 2016, have they really been getting better since their return? We can use show ratings on Phish.net to test this hypothesis, but let’s start by getting a few caveats out of the way.

Ratings on Phish .net are unlikely to be representative of the population of Phish fans as a whole. First, I suspect that .netters are among the most enthusiastic members of an already rabid fanbase. This may bias show ratings up or down--frankly, the hyper-critical nature of .netters means the direction of bias is not clear, but some sort of bias could be present. Second, not every .netter rates every show and I’m guessing that most of us are more likely to rate shows we’ve attended or watched on live video. This is, of course, the attendance bias effect discussed so often on .net. If attendance motivates a person to rate a show, then ratings on Phish.net will be biased upward. Third, a hot show on a current tour will sometimes zoom to the top of the all-time list, only to drop lower as more people provide a ratings weeks, months, or even years later. This “recency bias” results in higher ratings for recent shows relative to those from long ago. Fourth, the number of users on .net has been growing in recent years; if new .netters differ from long-time .netters (having, say, less familiarity with Phish performances from years past), then ratings from 2017 may not be comparable to those of 2009. Finally, the data reveal a “herd effect” for outstanding shows: more people will rate a great show than will rate poorly received show. We’ll be using the average rating for shows, though, so it’s not clear that the herd effect will bias a show rating. These caveats mean that our statistical analysis applies only to those of us on .net, and does not necessarily represent the broader population of Phish fans. That said, let’s take a look at the data.

Read more...