Beer with a confidence interval

Medalla is a Puerto Rican beer. On the side of a can it says

Alcohol by volume over 4%, not more than 6%.

I’d never seen a beer give a confidence interval for its alcohol content. I’d only seen point estimates before. For example, Budweiser announces that its beer contains 5% alcohol by volume. Just 5%, no statistical details.

I wonder why Medalla reports an interval.

  • Is their manufacturing process so variable that they can’t just report an average?
  • Is their manufacturing process no more variable than others, but they disclose their uncertainty?
  • Are they required to report alcohol content the way they do by local law or by the law of countries they export to?

Related posts:

11 thoughts on “Beer with a confidence interval

  1. I’ve been brewing beer for a year or so (at home, not commercially, the differences are significant). If I were to guess at the cause of the variability, I would say that they don’t have the ability (or are unable/don’t care to) measure/control the primary variables:
    fermentation temperature, mash temperature, water composition, and grain composition.

    Fermentation temperature has a tremendous impact on beer. Warmer fermentations will attenuate more (convert more sugar to alcohol/CO2) than cooler fermentations. A brief search for Medella beer indicates that it is a lager, so it will ferment for a long time at a cool temperature. I hope they would control this temperature carefully.

    The temperature of the mash is also very important. Usually you shoot for somewhere between 148 and 158 degrees. A lower temperature results in a highly fermentable wort, and higher temperatures result in a less fermentable wort. The sugars that don’t ferment are left in the beer, and are not turned into alcohol.

    Water composition is one of the most important variables in brewing. If the composition varies widely, and the brewery doesn’t filter and/or treat it, that may be a contributing factor. The pH of the water is a large determining factor is mash efficiency (how much starch in the grain you can convert to sugars relative to the ideal as measured by chemical methods by maltsters). Most brewers consider a pH of 5.2 to be ideal for most styles.

    Large breweries may also have well-equipped chem labs for analysis, which allows them to very precisely blend batches to produce a consistent product. Smaller breweries usually use third party services for these measurements, and may not have the resources to be nearly as precise. Many small breweries do not blend batches at all.

    Since grain is an agricultural product there can also be some variance in there as well. This is carefully measured by most maltsters (where the breweries will get their grains), and should be reported to the breweries so they can adjust accordingly.

    As a homebrewer, there are two ways to determine the alcohol content of your beer. You first measure the gravity (density relative to water) of your wort (unfermented beer) as it goes into the fermenter. You can do this with either a refractometer or a hydrometer. After fermentation completes (the yeast convert the fermentable sugars in the wort to CO2 and alcohol), you again measure your gravity. The difference between readings will give you the apparent attenuation (the amount of sugar converted to alcohol). Some of their variability may be attributed to imprecise gravity readings, although I doubt this is the case.

    In short, I would guess that most of that variability comes from shortcomings in measurement. I’ll leave the legal question to someone better informed than I am.

  2. So, the Second way of measuring alcohol content … does it involve a glass and one’s elbow?

  3. Chris,

    I’d guess your right with one of your reasons. My guess as to the cause of you being right would be ultimately money. They might have had to pay 100k for an expensive temperature control system but only 10k for a cheap one. So rather than go for consistency they saved the money and post a range.

    I suspect at some level, as with all food products, your variabliity needs to fall in a nice range for you to be able to say “50g” or 5% or whatever. They might have been required to post that range because they weren’t tight enough process controls to just say 5% legally.

  4. I’d guess it is for legal or tax reasons that they give the interval. In some jurisdictions there are higher taxes for higher alcohol content. I wonder if the interval is being used to avoid a higher tax rate.

    The EU stipulates:
    – Labeling must be accurate to within 0.3% for beer; 1.5% for wine; and 0.5% for spirits
    Not sure if you can use an interval and still claim to satisfy these requirements

  5. In general, beer manufacturers are very good at producing a consistent product from inconsistent ingredients. More on that here. If I had to guess, I’d say it’s more likely there’s a legal reason than a technical reason.

  6. By straddling 5% they thwart any claim that their beer is a statistically significant factor in anyone’s use or misuse of it 😉

  7. this really spoils my gossett/guiness story. I always try to motivate interest in stats by pointing out how important it is in brewing beer. with wine, people like variabilty, it’s part of the allure. With beer, you want to know that a Yorkshire Bitter is going to taste just the same next month as last year. How do you get such consitency – by understanding variability. How do you understand variabilty – with stats. but if you can’t get alcohol content that close, what hope is there for all the tannins (or whatever it is that regullates the flovour)

  8. Another possibility, by the way, is that it could be bottle conditioned. You can always tell exactly how much ethanol is going into the bottle, but if it’s bottle conditioned, you don’t necessarily know how much will be in there when you open it.

    Bottle conditioning is almost exclusive to smaller craft breweries and home brewers, of course.

  9. Actually folks have brewed beer for thousands of years without stats. To understand variability you need a controlled process. Stats are usually misapplied, misinterpreted and only necessary for legal or marketing reasons. Har har…

Leave a Reply

Your email address will not be published. Required fields are marked *