Cognitive Daily

i-eca0cf2af9fc3ac4445c7dff7d8aab70-research.gifThe developmental psychologist Jean Piaget developed several tasks to show how very young children were different from older kids. One of the most surprising is the “conservation” task: a 5-year-old, who talks clearly and appears quite bright, will watch water being poured from a short, wide glass into a tall, slender one. She will say that the tall glass has “more” water. A 7-year-old won’t make the same mistake.

Surely, then, adults are aware that short, fat glasses have deceptively large volumes, right? Not according to a recent article in BMJ. Brian Wansink and Koert van Ittersum asked 198 college students to pour shots of “liquor” (actually just dyed water) from a bottle into either short and wide or tall, slender glasses. They were allowed to practice pouring into a standard shot glass first. Here are the results:


The dotted line represents the volume of a shot. While more practice did reduce the volume of liquor poured, it did nothing to reduce the disparity between tall, wide, and short, slender glasses. College students, of legal drinking age, were making the same mistake as Piaget’s 5-year-olds.

College students, however, aren’t generally professional bartenders. Perhaps with years of experience, a bartender can accurately compensate for the width of a glass. Wansink and van Ittersum next offered actual bartenders $4 to try their skill at the task. These bartenders averaged 6.3 years’ experience. Here are their results:


Even bartenders made the same mistake as five-year-olds, pouring more liquid into the short, wide glasses than the tall, narrow ones. The bartenders weren’t offered practice trials, but they were more accurate than college students, and did even better when the experimenter told them to “please take your time” before pouring. However, even when taking care, they still poured significantly more liquor into short, wide glasses than tall, narrow ones.

Wansink and van Ittersum suggest that studies of alcohol consumption should take the shape of the glasses used into account when analyzing the data. Drinkers who favor tumblers over highball glasses may be consuming significantly more. And, if you’re trying to watch how much you drink, you might consider switching to taller, narrower glasses — or precisely measuring out each portion.

Note: The New York Times recently published an article highlighting more of Wansink’s research on how much we think we’re consuming versus what we really do consume. Fascinating stuff.

Wansink, B., & van Ittersum, K. (2005, 24-31 December). Shape of glass and amount of alcohol poured: Comparative study of effect of practice and concentration. BMJ, 331, 1512-1514.


  1. #1 MartinDH
    October 16, 2006

    I wonder what the errors would be when the experiment is run using glasses of many different widths. Do people estimate the quantity of liquor by its visible area (width of the glass * height of liquor) or do they attempt to estimate volume (w^2h/4)? What effect would using triangular or square glasses have?

  2. #2 Cathy
    October 16, 2006

    I don’t think the liquor pouring experiment relates to Piaget’s experiments in conservation. In Piaget’s case, he was watching for the children to recognize that pouring water from one vessel to another doesn’t change its volume. I doubt if any participants in the liquor pouring experiment thought the amount of liquor had changed just because it had been distributed among several glasses.

    This experiment does show that it’s extremely difficult to compare the volumes of differently shaped containers, and apparently that task doesn’t get much easier with age. That shouldn’t be too surprising. I would think most of us have had the experience of trying to figure out whether the dinner leftovers would fit in the smaller container or if we really needed the larger one. It is interesting that short wide containers look bigger than they really are.

  3. #3 Greg
    October 16, 2006

    What were they pouring with exactly? And what type of bartending background did these bartenders have? The reason I ask is that there are many things that can influence a pour. Having bartende in Las Vegas from 2002-2005, I learned a lot about pouring methods. Were pour spouts used? Were they plastic or metal/rubber? How much liquor (colored water) was in the bottles at the time of the pour? Are the bartenders freepour, jigger, or meter trained? Do they use the count method on a pour? Do they work in an environment where they get more tip for more liquor in the glass or a at a bar with gaming? Is a standard shot at their bar 1.5oz. or 2oz.? All of these variables come into consideration. I’m not trying to discredit the theory at all, just bringing up some things that could really affect how pours are done.

  4. #4 Dave Munger
    October 16, 2006


    The article (linked above) was rather sparse on the details. They were pouring from 1.5 liter bottles, and since there was no mention of pour spouts, I imagine they were not used. Shot size seemed to be 1.5 oz., though it wasn’t mentioned other than to say that bartenders attempted to pour that amount. No mention of the bartenders’ training.


    Good points. I think it does relate, but you’re right in that it’s a different task. Clearly adults understand the basic “conservation of volume” principle.

    Martin, I’d like to see that experiment too!

  5. #5 Daniel
    October 16, 2006

    I think the study is correct in it’s findings of volume perception, however the headline in this blog is incorrect. Most professional bartenders don’t measure pours by observation. If they are free-pouring they judge the amount of alcohol by counting the time of the pour.

  6. #6 Dave Munger
    October 16, 2006

    “If they are free-pouring they judge the amount of alcohol by counting the time of the pour.”

    Yes, but this requires a special spout in order to work. The point is that even bartenders can’t judge the volume accurately when they don’t have the proper apparatus.

    It’s also possible that bartenders might subtly vary the depending on the type of glass. I’d be interested to see this study retried, while allowing bartenders to use their special spouts.

  7. #7 Daniel
    October 16, 2006

    The issue I have with the headline is that the study wasn’t about pouring, it was about volume perception. It seems evident that the trial didn’t use the proper spouts for free-pouring or the line “please take your time” would make no sense in that context.

    Bartenders can and do pour accurately with the requisite tools.

  8. #8 Dave Munger
    October 17, 2006

    “Bartenders can and do pour accurately with the requisite tools.”

    I see your point. But college students could also pour accurately if they were allowed to measure the liquid in some way (using a shot glass, a special spout, a graduated cylinder, etc.). The point the headline was attempting to make is that bartenders make the same error estimating volume as college students while pouring from an ordinary bottle. I’m sorry if that was unclear.

    I do agree that it’s a little disingenuous of the study authors to suggest that in bars, glass shape is related to the amount of alcohol poured. Real bartenders measure the alcohol they pour; they don’t pour from ordinary bottles.

  9. #9 Casper Gielen
    October 20, 2006

    It seems obvious to me that estimating is easier in a tall glass.
    In a tall glass the variation in height of the liquid is much bigger.
    Let’s do some calculations.
    Let’s assume we are are pouring 10Pi units of booze. 10 Pi seems like a strange quantity, but it makes the calculations more easy.
    The tall glass has a radius of 1, the wide glass a radius of 2.

    r = radius
    h = height of the liquor

    Volume of a glass = Pi * r^2 * h = 10 Pi => r^2 * h = 10

    tall glass: 1^2 * h = 10 => h = 10
    wide glass: 2^2 * h = 10 => h = 2.5

    Now let’s add 1 Pi unit of booze.
    tall glass: 1^2 * h = 11 => h = 11
    wide glass: 2^2 * h = 11 => h = 2.75

    In other words, it’s 4 times harder to correctly estimate the amount if the glass is twice as wide.

New comments have been disabled.