Statistical Analysis of Beer Drinking

1. All the time, I'll often look up stats about the beer business just for my own curiosity or just to see where I am compared to the average beer geek/drinker.

2. If you compared me to an average beer geek in my market, I think my opinion and breakdown by style would be rather normal though with a much wider geographical footprint than most local beer geek. (Many of the beer geeks I interact with have an outsized love of local beer, there are definitely some winners, but I go for more of a variety from around the country)

3. I've wondered (but don't actually want to know):
-How much money I have spent on alcohol in my life

-An analysis of ratings compared to BAC at time of rating (I think we all get a bit more generous when we're half in the bag)

-Price elasticity at a personal level (I can tell you the price elasticity of any product in a macro sense, I suspect my sample size would be much too small and show that craft beer is less price elastic than toilet paper)

-An analysis comparing where I was drinking or with whom I was drinking and the standard deviation from the average rating here (for example, if I'm drinking at somewhere like Sierra Nevada Mills River, am I so blown away by the amazing surroundings that I give the beer a bump versus if I was drinking it alone in my basement? If I'm drinking alone, am I likely to pick a beer apart and rate lower versus when I'm kind of half focused on the beer and half focused on my conversation?

4. There's tons of value in the analysis of consumer trends. It'll be interesting to see if/when craft beer reaches a point when there is more analysis around craft brewery portfolios. The general problem with craft beer is simply that the sample size is rarely big enough to make strong conclusions. The follow up to that is "so what". If a brewery that has capacity for 5,000 bbl/year is selling every drop, there isn't much of a point in optimizing the business further because all you still can't make any more beer to sell.

I assure you, we already have it. IRI, retailer scan data, untappd data, guestmetrics (and others) for on premise, we have your loyalty card data, we have customer buying patterns from InfoScout... just for fun, I predicted what a buddy's wife would buy on her next trip to the store with 72% accuracy. It's really cool stuff!

 

I keep an Excel spreadsheet with the following information:

1. Brewery
2. State or country of brewery
3. Name of beer
4. Type of beer (e.g. IPA)
5. Sub-type of beer (if applicable)
6. Volume of the beer I drank in ounces. Obviously this is an estimate sometimes.
7. Alcohol %

I don't drink all that much, usually just one a day, so it's easy to remember exactly what I had.

I don't keep track of ratings. I also don't get too obsessive about recording every sip I take, for example if I had a few sips of someone else's beer I don't care about recording it. Also if I am doing samples I tend to lump them all together, without bothering to keep track of everything separately. For example when we went to Russian River we got those awesome sampler paddles, plus some pints that were shared. For that day I just tried to estimate how much beer I drank total. E.g. 48 oz. of beer, estimated blended % of 7.5%.

 

How many of 2,000 have you drunk more than one of?

 

Here's another interesting question to ponder with your stats as the beer industry grows, what state are many beers "from"? Is SNPA from California or North Carolina? Is Brewdog from Europe or Ohio (when they finish building), is Oskar Blues from Colorado or Texas? Is your Bud Light from Missouri, Colorado, Georgia, California, Ohio (etc.)

 

I hope I'm not the only one who went to the trouble to build a custom database for their ratings... wanted to have info on the breweries as well, and I refuse to flatfile where it's not appropriate. Plus I wanted to be able to be able to have multiple reviews for a given beer and, well, there's that flatfile thing again.

The sorting and etc. that we are able to do on this site is nice, but I really wanted to be able to slice and dice things, like "best tasting IPA from NC," "Nose of the Year," "highest ABV stout that still rates a 4 or above"... detailed stuff like that. Plus I use an aroma/flavor wheel as part of my tasting notes (I think it's helpful because it forces you to quantify) and I wanted to be able to use that data as well... "what are the hoppiest IPAs I tried this year," "darkest roast character in a stout," etc.

I've only done about 150 reviews so far, so I don't have a lot of data, but I've still managed to have a lot of fun with it.

Edit: Damn. I may need professional help.

 

My total is FAR below yours, @jrsaffert, but I do keep a spreadsheet. For each beer, it has date of the review, L-S-T-F-O rating, the BA rating, the RB rating, My personal formula rating, the brewery, the location of the brewery (state or country), the beer name, the style, the IBU, and ABV, and comments (that is, the text of the review).

Total beers in my spreadsheet: 102 (not all beers on BA or RB are in the spreadsheet; I started each at slightly different times.)
Average ABV: 6.1%
Average IBU: 44.4

Locations
Total states: 24
Most from single state: 62 (MN)
Total countries: 2, USA and UK ... this is all that is in my spreadsheet; before that, I have had beers from England, France, Belgium, Germany, Bavaria (I know it is part of Germany, like Texas is part of the USA) Austria, Japan, Mexico, Guatemala, Panama, Canada, and India; all of these while actually in those countries.

Styles
Number of different styles (using BA's style names): 37
Most rated style: American Pale Ale - 24 beers, avg rating 3.52
2nd most rated style: American IPA - 11 beers , avg rating 3.66
3rd most rated style: ESB - 6 beers, avg rating 4.13
Favorite beer style of all time (sadly, it is not in my spreadsheet and it has been a long time since I've had any): English Bitter, on cask.

This calculation is interesting to me, since I mostly prefer APAs to IPA. The explanation (I think) is that I tend to grab an APA I've never had before (hence, some not-so-good ones) whereas since IPA is not my favorite style, I tend to try them based on recommendation, etc. Also the IPA ratings are inflated by the fact that the fresh hop IPAs are included in that style.

Reviews (leaving out tick-ratings only - most of those were very early-on and are, frankly, not reliable for where I would rate them today)
Average BA review: 3.66
Highest BA review: 4.53 (Northgate Brewing Parapet ESB)
Lowest BA review: 2.36 (Surly 2015 reformulated Bitter Brewer)

 

I am attempting to work out a way to easily import data from the Your Beers sections into a spreadsheet without having to copy and paste everything individually. So far the formatting is making it difficult but I'll let you know if I have success.

 

0% chance I ever do this.

 

Oskar Blues also has a brewery in Brevard, N.C., just south and west of Asheville.

 

Price paid is another aspect I would want to throw into the mix if I ever do something like this. Is a $5 pint of fresh SNPA better than a $6/sixer SNPA. Essentially is the draft worth 3.7 times more than the cheaper bottle/canned alternative?

Likewise, does age or best by of said product get its own column too?

Crazy. I do in this a general sense on a price/value aspect (just to see). ABV & cost/oz is compared to a standard 50.4 unit normal beer.

So far the best value I have found is a Boatswain DIPA with a $0.5455/standard beer price. Don't laugh
SN Torpedo when its $12/12 pack, comes in at $0.583. Bud Light in a jumbo 30 pack, @ $0.70. People are overpaying if they just want to get drunk. Abyss and Southern Charred come in dead last among about 50 beers I have tried it with. They are at about $4/standard beer. Ten Fidy about $1.50. In fact most seem to sit right around $1.30-$2.00. Only when bombers come into play is the value aspect thrown way out of the window. Your BL $6/6 pack is your baseline $1.00/standard beer. What's interesting, many local growler options, priced at $10, are of the same inherent price value as bud light. Many growlers approaching or near that $1.00 mark. Why EVER drink swill?

I'd love to throw in some type of BA score multiplier or effect to add value to a standard price, and diminish the #s for the swill.


Name the best value:
Bud Light 30 pack @ 4.2%. $21
Local Imperial Java stout 6 pack @ 8%. $8

Answer. both. Identical. I guess drunks in general don't care to try to do any type of math.

Name the best value:
Bud Light 6 pack @ 4.2%. $6
FBS 4 pack @ 8.3%. $8

Answer. Both are identical. BMC drinkers scoff at the price. At least that's what I hear almost 80% of the time when I talk to swill drinkers.

 

So, it's pretty intuitive to do the basics but there are is a pretty large user community and a bunch of videos on tableau.com that will show you how to do some of the more esoteric stuff like the geo-mapping and using custom symbols (like the tulip glass).

 

Agreed - also, as you become exposed to more beers over time, palates evolve and horizons broaden - I think I'm going to systematically need to re-evaluate some of my earlier reviews to better rate to style. Would love to hear more about your mixed effects model when you get to that point!

 

Thanks! This can become an extension of my beer drinking "hobby"...

 

Anyone know if there is a way, or if BA is working on a way, to extract your own ratings, or anyone else's for that matter, into spreadsheet form so it is more easily sortable and searchable?

 

As another poster stated-550 beers from Minnesota???? Who'da thought that???
State with highest rated beers?
Low/High cost and average cost?
good work!

 

Few things that would be interesting with your ratings: a) standard deviation of your distribution from mean (brought up above by others), b) presence of skewness, positive or negative from your mean score (SKEW in Excel), c) kurtosis or fat-tails: abnormal number of very high or very low ratings (KURT in Excel). All easy enough in Excel once your ratings are laid out and the last one (c) especially would indicate something interesting: if you have very fat right tail, it means you're probably doing a great job in catching the most rewarding beers (unless the ratings are biased due to hype or something along those lines).

I suspect we're all catching more good beers than bad ones, in turn we have a selection bias towards beers we think will be more rewarding, so we have disproportionally high number of more highly rated beers, on average than a normal mean with normal distribution would indicate. However, some of us are generous graders, others grade more conservatively further making this be a noisy data set. Your mean of 3.87 indicates you are relatively more cool-headed than me, whereas my mean is probably closer to 4.0 as I tend to get overly excited. Our samples might be different as well, perhaps we're more choosy in the beers we pursue, affecting the means itself and distributions around it.

After teasing out info on your own beers, then you can compare your ratings vs. general ratings to expose additional layer of info on the deviation of your scores away from population average for the three figures listed above. I don't have a degree in stats or math, so apologies if anyone finds any inconsistencies or misinterpretations here. Excel makes this easy enough and this is more intuitive than most think. And yes, I think there is value to this analysis, we can tease our patterns, biases etc.

 

Ok, I did put some statistics together.
2,800 beer reviews, place or origin. USA 2,154 76.9%
Out of 49 states and DC reviewed. (ND is the only one missing)States with at least 100 beers reviewed:
PA 292 10.4%
CA 193 6.9%
NY 164 5.9%
CO 157 5.6%
WI 140 5.0%
MD 117 4.2%

Out of 76 countries with a beer review, the top 5 other than the USA:
Germany 90 3.2%
Belgium 59 2.1%
Canada 48 1.7%
England 47 1.7%
Mexico 32 1.1%

Beer Styles. Out of 99 styles reviewed, the top 10:
American IPA 292 10.4%
American Adjunct Lager 159 5.7%
American Pale Ale 136 4.9%
Euro Pale Lager 106 3.8%
American Amber/Red Ale 99 3.5%
Fruit/Vegetable Beer 82 2.9%
Witbier 79 2.8%
American Pale Wheat Ale 77 2.8%
German Pilsener 73 2.6%
American Double IPA 72 2.6%

 

You guys should also try ranking styles by your highest to lowest rated.

 

There is good news. There is probably no limit to these analytical ponderings to track down, quantify, and know.