Attribute Framed

A glass half full is also half empty.

Posted Oct 02, 2020

J. Krueger
Attribute framed
Source: J. Krueger

The consumers' evaluations were more favorable toward the beef labeled “75% lean” than that labeled “25% fat.” —Levin & Gaeth, 1988, p. 374

Attribute framing is a type of context effect that is often said to reveal underlying irrationalities. Consumers, and people in general, should, the argument goes, attend to the underlying quantitative information and ignore how it is presented. Frames are chosen by agents with economic interests and the consumers pay the price. Presumably, meat-eaters are more tolerant of fat intake when the frame refers to leanness instead of fattiness.

According to the irrationality view, you would — but should not — choose differently given a different frame, and when the choice of frame is not up to you, you are exploitable. Don’t blame the framers. Why should they not choose the frame that makes it more likely for you to do what they want you to do? To be rational, consumers would have to generate the not-presented frame from the provided information, realize that there is no logical difference between this and the presented frame, and then ask themselves whether they still want to eat beef that contains 25% fat and 75% other stuff.

The irrationality argument overlooks that in many instances, the presented frame carries information beyond what is stated explicitly and numerically. Being told that the glass is half full you might infer that the glass is in the process of being filled; if you wait a little, you’ll have more to drink. Being told that the glass is half empty you might infer that the glass is in the process of draining out. If you wait a little, you’ll have less to drink. So better have a drink now. Ditto for the bathtub — without the drinking. These inferences about surplus information can be quite rational in the context of well-behaved conversations (Grice, 1975). But advertisers can exploit the healthy inclinations of the pragmatic mind and assume that beef eaters will prefer 75% leanness over 25% fat, while not giving themselves a full account of the fact that the beef’s leanness is not increasing.

Being taken in by frames is a kind of innumeracy (Peters, 2020), at least when there is no adaptive advantage to assume that the framers meant to imply additional information that is good for you. The hapless consumer today is overwhelmed with numerical information on food items. Some of this information must be provided by law, whereas other information is simply meant to bamboozle you. One technique to overwhelm the consumer is to present so many numbers that a fine-grained search is needed to find the value of interest — and who has the time? Another technique is to corrupt the numerical information to the point of incomprehensibility.

Looking for cranberry juice at the corner store, I found a product said to “contain 100% juice.” What does this mean? It does not say that the bottle’s contents are 100% juice. Instead, it seems to hint that whatever juice is contained in the mix, that is 100% juice. Another bottle bore the exciting news that it contained a “100% cranberry juice blend.” This message too qualifies for the nonsense files. Let’s say X-percent of the contents are 100% juice blend, then what is the composition of the 100 — X part of the drink?

Undeterred, I continued my quest for pure juice and spotted a bottle said to contain “100% cranberry juice.” Now this, dear friends, is more like it, I thought. There were two clues that I had found what I was looking for. One was the bottle’s price, which was high. The second, post-purchase, clue was the intensely tart taste. I had to — as I had hoped — do my own diluting, which I was able to accomplish with water and 100% no sugar.

Having scored points as a wellness- and numeracy-minded consumer, I thought I might reward myself with a minty treat. The wrapper holding my object of desire cheerfully announced that the chocolate mint contained “70% less fat.” "Hooray," says the fast-thinking System 1, but "What does this mean?" asks the slow-reasoning System 2 (Kahneman, 2011). The back of the wrapper reveals that there are 2.5 grams of fat, and this suggests we used to get 8.4 grams with each treat. The unintended side message is this: "We used to sell you fatty treats, but we have since reformed ourselves."

Not knowing the total weight of the treat without weighing it, I wondered if I could use the provided information to create an alternate frame. What is the equivalent percentage increase in fat-freeness? Let’s take the 8.4 grams of fat we used to get and assume that the treat’s total weight is 100 grams. This means we got 8.4% fat and we are now down to 2.5%. If we were to say the treat is now 30% more fat-free, we’d be saying that we have moved from 91.6 percent fat-free (100 - 8.4 = 91.6) to 119.6% (91.6 + .3 x 91.6), which is not possible.

Now suppose the total weight of the treat was 36 grams. With 8.4 grams of fat, 76.67% would be fat-free, and an increase of 30% in fat-freedom would make the whole treat fat-free. The math is beginning to work but then breaks down. Where did the 8.4 grams of fat go? Preserving the claim of a 70% drop of fat contents from 8.4 grams to 2.5, and assuming a total weight of 36 grams, we can have an increase in fat freedom of 21.4%, going from 27.6 grams nonfat to 33.5 nonfat. This works out numerically, but who would want to compute it? 

Students bring their own frames to the table when thinking (and complaining) about grades. They seem to think that they had 100% going in, and then they demand to know where they lost 10% if they only got 90%. I have often offered them an alternative frame, built on the assumption that they had 0% going into the exam so that the points received are points gained. My attempt to leverage the bathtub analogy (points/water have been added) is rarely well-received. I am becoming more sympathetic to the students’ framing because, with grade inflation, easy exams, and talented students, most of the numerical variation in exam percentages is much closer to the 100% mark than the 0% mark. This distributional skew directs attention to what's missing and away from what’s being had. Perhaps the students’ focus on this one frame is not completely rational, but it is all too human.

Note. As to this post's byline, we might also say that "A glass half empty is also half full." It changes the outlook. It's a tribute to frames.


Grice HP (1975) Logic and conversation. In P. Cole & J. L. Morgan (eds.), Syntax & Semantics, Vol. 3. Academic Press.  

Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux.

Krueger, J. I. (2020). The innumerati. Psychology Today Online.

Levin, I. P., & Gaeth, G. J. (1988). How consumers are affected by the framing of attribute information before and after consuming the product. Journal of Consumer Research, 15, 374–378.

Peters, E. (2020). Innumeracy in the wild. Oxford University Press.