I’m not a statistician, so this is a bit of a stretch for me. But I’m posting this information because the internet is smarter than I am.

Last week, in the discussion of my post about what certain numbers (Specific Gravity, IBU, etc.) mean about your beer, the idea came up to make what amounts to a simplified statistic about beer – a way to represent, more simply, what people really want to know about their beer. It goes along with the pipe dream of having breweries print statistics about their beer on the labels.

The theory, in my mind, is that nobody wants to pick up a beer and think, “1.056 and 1.010 with 60 IBUs.. hrm… that’s pretty bitter.” Some beer geeks can do that in their heads, but a lot of people can’t, or don’t want to. On the other hand, if you could pick a beer and say, “A hoppiness rating of 92? Good heavens! That’s big!” it would be pretty cool.

So, I’ve been playing with numbers a wee bit. Mostly collecting them. I started with the BJCP Style Guidelines. I recorded the maximum and minimum stats for OG, FG, and IBUs and calculated what the average beer would look like in each style category.

For example, for English IPA the style range shows:

OG: 1.050 – 1.075
FG: 1.010 – 1.018
IBU: 40 – 60

Thus, the average English IPA would be:

OG: 1.063
FG: 1.014
IBU: 50

This is just to give a basic sample range of beers to work with that should be fairly representative of all the styles involved, though I’ve considered removing sour styles from this exercise, since Lactic Acid kind of fouls everything up on the whole, “This is what flavor you should expect” front.

I also calculated apparent attenuation for each style – finding out how much sugar has been fermented out from each on average.

(GU – FU)/GU

The Average English IPA listed above has an Apparent Attenuation of 77.60%. In other words, 77.6% of the sugar in the solution has been converted to alcohol. In reality, that’s not quite right, since alcohol is lighter than water and this is being calculated by the density of the liquid. You can ferment down to a final gravity lower than 1.000. However, for this purpose, this calculation should be good enough.

Then, I made what I’m (currently) referring to as a “hoppiness score.” It’s on the same line as a GU:BU ratio, but instead what I did is divide IBU by GU, then multiply it by 100 to give us a nice round number that we can related to instead of a decimal. The theory is that the more hops there are in comparison to original gravity, the higher this number would be. It should correspond with how hoppy the beer is. It works fairly well.

The English IPA up there would have a hoppiness score of 80. Is it an arbitrary number? Sure. Work in progress. Bear with me.

Then I applied apparent attenuation. It was noted in the discussion that the more dry a beer is, the more bitterness would be apparent to the drinker. I agree. To account for this, I multiplied the hoppiness score by the apparent attenuation. I called this “Apparent Bitterness.” I figured that the drier a beer was, the higher the apparent attenuation would be and so the closer the “Apparent Bitterness” would be to the Hoppiness Score. In the full list of styles, it does what I was hoping for and tends to give styles in which you would expect more bitterness a higher score.

Here’s the list I was working from sorted by Hoppiness Score:

And the same chart sorted by Apparent Bitterness:

It’s not quite right, but it’s definitely headed in the right direction. I would love to hear from anybody who has different ideas on how to represent these numbers, and you can damn well bet that I’ll be posting more as I fiddle around with math.

Maltiness coming soon, it’s a lot more challenging.

[Head on over to Part 2]

Earlier this week, I had my first try of Westvleteren 12, the so-called best beer in the world. No doubt, it was awesome; indescribably wonderful. When I checked later on, though, I noticed that while it was listed #1 at Beer Advocate, it was listed #2 at Rate Beer. Interesting.

It got me to thinking about the differences between the two sites and how much they agreed with one another. I started to take a closer look at what was listed at both sites.

As Andy Crouch noted earlier this week, there is a distinct lack of lagers on each of these lists, and an abundance of barrel-aged and/or hop heavy and/or alcohol heavy offerings. They’re also both heavy in rare, small-run, and hard-to-find beers. I suppose it’s all very American. Bigger is better and if it’s hard to get it must be awesome. Sounds like a recipe for eBay, if you ask me. But that’s not my focus today. That’s for my “please make more session beer” column later.

What I found fascinating was the agreement between the two lists. First of all, I found it interesting that more than half of the beers appearing on one list do not appear on the other (52). In Rate Beer’s case, 6 of the Top 10 beers they have listed do not appear in Beer Advocate’s Top 100 whatsoever. Only 1 of Beer Advocate’s Top 10 does not appear in Rate Beer’s Top 100.

So I cut myself down to looking at only the 48 beers that appear in both lists. Of those 48 beers, there is very little close agreement. Only one matches right on. Pizza Port Cuvee de Tomme ranks at #95 on both lists. The next closest agreement is the aforementioned Westvleteren 12. Only 27% of the list (13 out of 48) were in what I would consider close agreement (within 5 places, plus or minus, of the other list), whereas 38% of the list (18 out of 48) were more than 20 places apart.

I also threw a couple of scatter plots together.

They’re both the same scatter plot, sorted two different ways. Scatter 1 is sorted by BA rank (thus stripe of blue up the middle) and Scatter 2 is sorted by RB rank (thus the strip of red up the middle). You can see from these that, of the beers that both sites ranked in the Top 100, Beer Advocate tended to rank the beers higher (lower in number: Rank 1 = The Best).

The number of times that the following words appear in both lists combined (if a beer appears on both lists, the word was counted once):

Bourbon: 10
Barrel: 16
Aged: 15
Imperial: 19
Stout: 31
Ale: 10
IPA/India Pale Ale: 7
Black: 7
Hop/Hoppy/Hoppiness, etc: 6
The suffix “-ation”: 8
Lager: 0

You’d almost think that stouts, and especially bourbon barrel aged ones were the most popular craft beers on the market, and not IPAs.

What final conclusion can we draw from all of this? It’s hard to say. Since they have two different ranking systems (5 point scale vs. 100 point scale) it’s difficult to draw any specific comparisons. Mostly, it’s an interesting look at the tastes of the user base at both sites. I wonder how many people rate at both sites and how their ratings compare given the different point systems.

I also put both lists together (where the beers match) and came up with a mean average of scores to give the overall Top 48 beers. Here’s the list: