Medium Maximization

What it is

A medium sits between effort and reward. A person does a task, receives points, and later trades the points for something they want. The trade is a separate step, and the idea in Hsee's paper is that people can begin to judge the options in front of them on the scale of the points instead of on the scale of what the points will buy.

The authors argue that the medium works through three illusions. It can give an illusion of advantage to an option that is otherwise not so advantageous. It can give an illusion of certainty to an option that is otherwise uncertain. And it can give an illusion of linearity to an effort and outcome relationship that is in fact concave, meaning that each extra unit of effort buys less than the one before.

The name signals that what gets maximised is the medium, not the outcome. It is a claim about what gets counted at the moment of choice. It is not a claim that people value the points more than the reward, and the paper does not say that people never redeem them.

The finding matters for loyalty programmes because a programme sets the scale. It decides how many points a purchase earns and how many a reward costs, and so it decides the numbers that the holder compares. That is a reading of the paper's mechanism applied to a setting the paper names as an example, not a result the paper tested.

In effect

The experiment that carries the idea is the first in the paper. Ninety-six students were told they could do one of two tasks for a gallon of Haagen-Dazs ice cream. The short task took six minutes and led to vanilla. The long task took seven minutes and led to pistachio. For one group that was the whole offer. For the other group the tasks paid points, sixty for the short task and one hundred for the long one, and the rule was that any total from fifty to ninety-nine points bought vanilla while one hundred bought pistachio. A larger share took the long task when points stood in the way. Liking for pistachio and vanilla did not differ between the two groups, so the points had not changed what anyone wanted.

In commerce, the same structure is every programme in which the unit people collect is not the unit they spend. Hsee's team name frequent-flyer miles redeemable for travel and loyalty points redeemable for gifts, and a school token economy in which chips are traded for sweets. In each, the holder is asked to value a number first and a reward second.

In everyday life it shows up whenever a person compares two rewards by their point prices before asking what either one would actually buy. This is a plain illustration of the mechanism, not a documented case.

What it does not say

It does not say that people cannot redeem points, or that holding points is a mistake. The experiment asked students to choose between two tasks. It did not measure when anyone redeemed anything, so it says nothing directly about why points sit unspent.

It rests on a laboratory choice with a small paid reward. For this page the abstract and the text of the first study were read, and the paper's other studies were not. The exact shares choosing each task are given in a figure that was not available as text, so they are not reported here.

The link from medium maximization to the unspent balances studied in loyalty programmes is an inference. The research on unspent balances, such as Stourm, Bradlow and Fader's, points to separate mental accounts and the effort of redeeming, and it does not test this mechanism.

How it differs from mental accounting

Medium maximization and mental accounting are close neighbours, and a loyalty balance can involve both. Mental accounting, in Richard Thaler's account, is about where money is booked: people sort it into separate categories and judge each transaction inside its own category. Medium maximization is about what is counted at the moment of choice: the number of points, and not what the points will buy. A holder who keeps points in a separate account from cash is doing the first. A holder who compares two rewards by their point prices before asking what either would buy is doing the second.

The arithmetic of the ice cream experiment shows the scale effect plainly. On the points scale, the long task paid 100 points against 60, two thirds more, for one extra minute of work, which is about a sixth more time. The gap on the points scale looked large, and that is the sense in which the medium gave the longer task an illusion of advantage.

---