Analyzing 1,570 IPAs from Tree House, Trillium, Other Half and Monkish

Note: The raw data from December 14, 2020 can be downloaded here.

Rikunert and Koen Van den Heuvel both previously demonstrated there is a correlation between alcohol by volume (ABV) and average rating/score (on Untappd and/or BeerAdvocate). Curious to see if this was true for New England IPA (NEIPA), I transcribed every Tree House, Trillium, Other Half, and Monkish IPA on BeerAdvocate into Excel. After removing incomplete samples, I analyzed the remaining 1,570 IPAs using Excel:

Using the combined data set, I calculated the normal distribution and graphed the data (see the figure above). Note:

• Solid blue line = normal distribution of ABV %
• Solid dark green line = mean
• Dashed dark green lines = standard deviations
• Solid red lines = lower/upper range limit
• Solid red area = outliers
• Dashed light green lines = Brewers Association (BA) Style Guideline ABV % for Juicy/Hazy India Pale Ale and Juicy/Hazy Imperial/Double India Pale Ale
• Solid light green lines = mean for beers within Brewers Association Style Guidelines

Furthermore, I fitted the data sets to a linear regression analysis and calculated the coefficient of determination (R²) to determine how closely the model fits the data. The R² results were:

• Combined = 0.0718
• Tree House= 0.0047
• Trillium = 0.00007
• Other Half = 0.0005
• Monish = 0.0004

Based on the low R² values, there is a very weak correlation between ABV and average rating/score for these 1,570 IPAs. In other words, higher NEIPA ABV does not produce higher rating/scores.

But the analyses above do suggest a good target for NEIPA ABV is between 6.89 and 9.20%. This range includes the overall mean (8.04%), BA NEIPA mean (6.94%) and BA NEDIPA mean (8.57%) and is bound by one standard deviation above and below the overall mean.