011 min
Why a reference point changes how a price feels
Say a designer at a subscription app has to show a new price: $12 a month, up from $9. The numbers here are made up, but the pattern behind them is well documented.
Two people open the same page. One has paid $9 for a year. The other is new and sees the $12 plan next to a $20 plan. The price on the screen is the same for both of them. Their reactions are not.
The current subscriber sees $12 and thinks of the $9 they pay now, so the page asks them to give up $3 a month. The new visitor sees $12 and thinks of the $20 beside it, so the page offers them $8 off.
Neither person reacts to the number 12 itself. Each reacts to the distance between 12 and the number they compare it with. That comparison number is the reference point, and it explains many choices that look odd when only the amounts are looked at. A designer controls part of it, because the page decides what sits beside the price. The rest of this article follows the $12 plan to see how far that control goes and where it stops.
021 min
How the reference point turns a price into a gain or a loss
The difference between the two readers comes from how outcomes are coded. Kahneman and Tversky showed in 1979 that people usually see an outcome as a gain or a loss, not as a final total of wealth. Researchers call this seeing "coding". A gain or a loss is measured from a neutral starting point, and that starting point is the reference point. For the two customers in the $12 example, the coding works out like this.
| Person | Reference point | How the $12 plan is coded |
|---|---|---|
| Current subscriber | The $9 paid now | A loss of $3 a month |
| New visitor | The $20 plan beside it | A saving of $8 a month |
The coding matters because gains and losses do not weigh the same. A loss counts for more than a gain of the same size, a pattern called loss aversion. The subscriber's $3 therefore weighs more than a $3 saving would. Sensitivity also falls with distance, so each extra unit of change counts for less the further it lies from the reference point.
The coding changes how people treat risk as well. Kahneman and Tversky found that people tend to avoid risk when a sure gain is on offer and to accept it when they face a sure loss.
031 min
Where the reference point idea comes from
The reference point idea has a clear starting date. Before 1979, the standard way to model a risky choice was expected utility theory, which values each option by the total wealth it would leave a person with. Daniel Kahneman and Amos Tversky published "Prospect Theory: An Analysis of Decision under Risk" in the journal Econometrica in March 1979. They argued that value attaches to changes from a starting position, not to totals, because people respond to what they gain or lose. In 1992 Tversky and Kahneman revised the model and named the revision cumulative prospect theory.
The idea comes from how the senses adapt
The authors did not invent the starting position from nothing. They borrowed it from perception. Eyes, ears and skin respond to differences, so the same water can feel hot to a hand that was in cold water and cold to a hand that was in hot water. The level the sense has adapted to acts as the reference point. Kahneman and Tversky argued that money, health and prestige work the same way. The same wealth can mean poverty for one person and riches for another, depending on what each already has.
042 min
What sets the reference point
Kahneman and Tversky's model needs a reference point but does not say exactly where it comes from. The research points to several sources, and they can disagree inside one person.
The current position
In most of the examples in the 1979 paper, the reference point is the position the person holds now. The subscriber's $9 is the plain case.
Recent changes not yet absorbed
The reference point can lag behind the facts. The authors describe a person who has recently lost money and has not yet absorbed the loss, so they still measure their choices from the old position. A subscriber whose bill rose last week may likewise keep comparing every offer with $9.
Expectations
Expectations can set it too. The 1979 paper gives the case of an unexpected tax withdrawal from a monthly pay check, which people experience as a loss and not as a smaller gain. In 2004 Botond Kőszegi and Matthew Rabin went further, in a paper that appeared in the Quarterly Journal of Economics in 2006. They proposed that the reference point is a person's recent expectations about outcomes, not the status quo.
Their model has a direct consequence for pricing. The lower the price a buyer expects to pay, the more a higher price feels like a loss, and the less the buyer will pay. A visitor who has already seen the same plan for $9 elsewhere arrives expecting $9, so $12 is coded as a loss from the start.
Targets and other people
A target is a reference point as well. The next section follows par in golf, where the scores of competitors also come into play.
What the page puts beside the number
A designer has the most control over the last source. The 1979 paper says the way options are presented can change where the reference point sits. A price shown next to an earlier price, a plan shown next to a dearer plan, or a result shown next to a target each hands the reader a comparison. Presenting the same facts with different comparisons changes choices, a pattern called the framing effect.
052 min
Par as a reference point in professional golf
Professional golf gives a clean test of the reference point, because the reference number is printed on every hole. A golfer is paid for total strokes over a tournament, yet each hole has a par, the number of strokes an expert is expected to need. A birdie putt, one stroke under par, is a chance to gain against par. A par putt is a chance to avoid losing against it. If only the total mattered, both putts would be equally important.
Devin Pope and Maurice Schweitzer studied more than 2.5 million putts on the PGA Tour, each measured by laser. From the same distance, golfers made birdie putts about 2 percentage points less often than par putts. The authors read the gap as extra focus on par putts, because a miss there is coded as a loss.
The way golfers missed fits the same story. When golfers missed, they hit birdie putts less hard than par putts, and more of the birdie misses stopped short of the hole.
The gap is not fixed
The accuracy gap was largest in the first round of a tournament and less than half as large in the fourth. Pope and Schweitzer suggest that par stands out early in a tournament, while later the scores of competitors stand out more. If that is right, the reference point moved during the event. One decision can therefore have a different reference point at different moments, and several references can compete for the same decision.
061 min
Raising a price: the same $200 read two ways
Researchers have tested how a reference point shapes judgments of a price rise. In 1986 Daniel Kahneman, Jack Knetsch and Richard Thaler reported telephone surveys in which adults in Toronto and Vancouver rated whether business actions were fair. A hardware store that raised the price of a snow shovel from $15 to $20 after a snowstorm was called unfair by 82 percent of respondents. The authors explained the result with a reference transaction, the latest price a customer has come to expect. Customers feel entitled to the terms of that transaction.
Another pair of questions is closer to the $12 plan. Say a popular car model is in short supply. In the first, a dealer sells it at list price and then charges $200 more. In the second, a dealer sells it at $200 below list price and then charges list price, so customers also pay $200 more than before. In the first case 71 percent of respondents called the change unfair. In the second, 42 percent did.
The authors read the difference as a difference in reference points. The first increase is a loss against a clear reference, the list price. In the second the reference is ambiguous, and some respondents saw the change as the end of a gain. The dollar amount was identical, so the judgments moved with the reference and not with the amount.
071 min
How the reference point shows up in interfaces
Interfaces give people reference points all the time, often without anyone deciding to. Each item below is a design choice about what sits beside a number.
- An earlier or higher price shown beside the current one. The reader gets a comparison. Price anchoring covers how sellers use this.
- Scores reported relative to par. Golf tournament scores are totals of each round's score relative to par, so the reference is part of the number itself.
- A pre-selected plan or setting. A default can become the point from which other options are judged, which is one reason the default effect matters in forms and pricing tables.
- A change figure such as "since last month". It makes last month the reference.
- A goal marker on a progress display. It makes the target the reference.
- An earlier notice. In 2013 Idris Adjerid, Alessandro Acquisti, Laura Brandimarte and George Loewenstein studied how a privacy notice a person saw earlier shapes their response to a later one.
When a reference point is chosen with care, it helps people read a number correctly. The next two sections cover two cases where it needs to be removed or must not be invented.
081 min
When counting totals removes the reference point
A reference point can be removed by changing how the choice is counted. Kahneman and Tversky noted that a person who frames a choice in final totals, not in gains and losses, sets the reference point to zero wealth. In their analysis, that way of counting largely removes the habit of taking risks to avoid a loss.
For the $12 plan, a total frame would say that a year now costs $144 against $108 before, a difference of $36, with no word about a monthly rise. A design that shows whole-period totals pushes readers toward counting that way. No study cited here has tested that on a pricing page, so treat it as an idea to test.
091 min
Made-up reference prices
A crossed-out earlier price works as a comparison only if the earlier price was real. The United States Federal Trade Commission has a rule on this for advertisers, in its Guides Against Deceptive Pricing. A former price can support a price comparison if it was the real price at which the item was offered to the public regularly, over a reasonably substantial period.
Other countries have their own rules, so check the market you sell in. A designer who adds a crossed-out price that nobody ever paid is not choosing a reference point. They are inventing one, and readers who notice stop trusting the page.
102 min
Reference point versus nearby concepts
Several neighbouring ideas describe related effects, and they are easy to mix up. One axis separates them: what each one changes about a judgment.
| Concept | What it describes | How it differs from a reference point |
|---|---|---|
| PsychologyAnchoring | ||
| MarketingPrice Anchoring | ||
| PsychologyLoss Aversion | ||
| PsychologyEndowment Effect | ||
| PsychologyStatus Quo Bias |
A reference point is not an anchor
The two terms are often mixed up. An anchor is a number that pulls a later estimate toward it. A reference point is the zero of the scale on which gains and losses are counted. In Kahneman and Tversky's model, the value function measures how far an outcome is from the reference point. A crossed-out price can do both jobs at once, which is why the terms blur. To tell them apart, ask whether the number moved a guess or turned an outcome from a gain into a loss.
111 min
Where the evidence on reference points is contested
Researchers agree that people judge outcomes against some comparison. They disagree about three things.
Where the reference point sits
Locating the reference point is the hardest practical problem. Kahneman and Tversky called its location a critical factor in the analysis of decisions, and Kőszegi and Rabin said their own model still needs psychological and economic judgment about what counts as a recent expectation. Kőszegi and Rabin also argued that expectations predict better than the status quo where people do not expect to keep what they have.
Whether losses outweigh gains
David Gal and Derek Rucker reviewed the evidence in 2018 and concluded that it does not show losses, on balance, having more impact than gains. Replies published in the same journal argued for a narrower claim, that losses outweigh gains in some settings. Psychologists have also objected that prospect theory describes choices without explaining the mental process behind them.
What this leaves a designer
A designer cannot read a person's reference point off the screen. Treat it as a guess to test. In user tests, ask people what they compared a price or result with, and compare reactions to versions that show different comparisons.
?3 questions
Questions people ask
Can a person use more than one reference point at once?
Does a reference point have to be a number?
§11 sources
Sources
Kahneman, D. and Tversky, A. (1979). Prospect Theory: An Analysis of Decision under Risk. Econometrica 47(2).
Kőszegi, B. and Rabin, M. (2004 working paper; published 2006 in the Quarterly Journal of Economics). A Model of Reference-Dependent Preferences.
Pope, D. and Schweitzer, M. (2011). Is Tiger Woods Loss Averse? Persistent Bias in the Face of Experience, Competition, and High Stakes. American Economic Review 101(1).
Pope, D. and Schweitzer, M. (2011). Author copy of the same paper.
Show all 11 sourcesShow fewer sources
Kahneman, D., Knetsch, J. and Thaler, R. (1986). Fairness as a Constraint on Profit Seeking: Entitlements in the Market. American Economic Review 76(4).
US Federal Trade Commission. 16 CFR 233.1, Former price comparisons.
Tversky, A. and Kahneman, D. (1992). Advances in prospect theory: Cumulative representation of uncertainty. Journal of Risk and Uncertainty 5.
Gal, D. and Rucker, D. (2018). The Loss of Loss Aversion: Will It Loom Larger Than Its Gain? Journal of Consumer Psychology 28(3).
Adjerid, I., Acquisti, A., Brandimarte, L. and Loewenstein, G. (2013). Sleights of privacy: Framing, disclosures, and the limits of transparency. Symposium on Usable Privacy and Security.
Wikipedia, Par (score): how golf scores are reported relative to par. (score)
Wikipedia, Reference dependence: a summary that includes the marathon study by Markle and colleagues.
