Framing: Why the Same Choice Reads Differently

A frame is the description that fixes what counts as a gain and what counts as a loss. Change it and the same outcome, at the same odds and the same amounts, draws a different choice out of the same person. In the invented decision log that swing is 35.0 points wide. No description is neutral. The absence of a neutral description is the uncomfortable half of the idea.

The whole effect rests on one property of the way people judge outcomes. Almost nobody judges a final position. People judge a change. Asked how the month went, a person does not report the balance in the account; the report is whether it went up or down, and by how much. The habit of judging changes is so ordinary it takes an effort to notice at all. The moment a judgement is about a change rather than a final position, it needs something to be a change away from, and that something has to be supplied by somebody. In an experiment it is supplied by the person writing the question. In ordinary life it is supplied by whoever wrote the sentence that happened to be read.

Take it out of finance before taking it into finance. A vegetable seller on a street corner counts Rs 1,400/- at the end of a day. Set that against yesterday, when Rs 1,600/- came in, and the day was bad by Rs 200/-. Set it against the Rs 1,200/- he has to take to cover the stall and the stock, and the day was good by Rs 200/-. The Rs 1,400/- did not move. The number that moved was the one sitting quietly behind the sentence, and that unstated starting point is what a frame supplies.

What is a frame, and what does changing one actually change?

A frameA description of an outcome that fixes what counts as a gain and what counts as a loss. is a description of an outcome that fixes what counts as a gain and what counts as a loss. A frame is not spin, it is not exaggeration and it does not require anything in the sentence to be untrue. Every word in a framed statement can be exactly correct. The frame is doing its work one level below the truth of the sentence, at the level of the starting point the sentence quietly measures from.

So list what changing a frame does not change. The outcome stays whatever it is. The odds attached to that outcome stay where they were. Every amount of money involved stays where it was. Anything a careful person would compute comes out the same. Changing a frame changes nothing about the outcome and everything about the point the outcome is counted as a movement away from. The change is small to make and large to receive.

Look at one position in the invented Palash decision log to see the structure without any drama in it. The Nilgiri mid-cap scheme cost Rs 3,00,000/- and is worth Rs 2,55,000/- when the eight-quarter valuation is struck on 30 September. Both sentences that follow are true. The holding is down Rs 45,000/-, being 15.0 per cent. The holding is worth Rs 2,55,000/-. The first sentence measures from the purchase cost. The second measures from nothing at all. Neither one is lying, and a reader who takes them in on different days will not react to them the same way.

What happens when the same outcome is described two ways?

The invented Palash decision log records a measurement taken on the 60 investors in it, and the measurement was built for exactly this purpose. Each person was offered a straight pair. Either take Rs 5,000/- for certain, or take a half chance of Rs 11,000/- and a half chance of nothing. The first offer stops there. The second offer is the same pair with every amount and every probability held where it was, and one thing altered: the outcome arrives as something given up rather than something received. Either give up Rs 5,000/- for certain, or take a half chance of giving up Rs 11,000/- and a half chance of giving up nothing.

Set the two offers beside each other and count what differs. The certain optionThe choice with one single known outcome, rather than a spread of outcomes with probabilities attached. is Rs 5,000/- in both. The gamble is a half chance of Rs 11,000/- in both. The expected valueThe average outcome, weighting each possibility by how likely it is. Half of Rs 11,000/- is Rs 5,500/-. of the gamble is Rs 5,500/- in both. Half of Rs 11,000/- is Rs 5,500/- whichever direction the money is travelling. The only difference between the two offers is the pair of verbs, received against given upDescribed as a loss counted down from a higher starting point, rather than as a gain counted up from a lower one., and everything a calculator could touch is identical.

In the first wording, 42 of the 60 took the certain Rs 5,000/- rather than the gamble. Forty two of sixty works out at 70.0 per cent. The result is comfortable and recognisable, and nothing about it is surprising: most people, offered a sure amount against a coin toss, take the sure amount even when the coin toss averages more. Seventy per cent is the number to hold against the second wording, where the identical structure arrives as something given up.

How large is the reversal when it is measured rather than asserted?

Here is the second reading in full. Offered the certain Rs 5,000/- loss against the half chance of losing Rs 11,000/-, 39 of the 60 took the gamble. Sixty less thirty nine leaves 21, so 21 of 60 took the certain loss. Twenty one of sixty is 35.0 per cent. Put the two readings together and the share taking the certain option went from 70.0 per cent to 35.0 per cent, a fall of 35.0 points and a little over half the room changing sides.

Seventy per cent took the certain amount when it arrived as a gain and thirty five per cent took it when the identical structure arrived as a loss, and both readings come from one group of sixty people on one afternoon. One group of sixty people on one afternoon is what makes this a measurement rather than a debating point. The reading is not one group compared with another group, not one year compared with another year, and not a survey compared with a different survey. Same people, same room, minutes apart, with a reversalThe same person choosing differently when the same outcome is described in a different way. that is wider than most effects anybody bothers to name.

Is the framing effect a mistake in the arithmetic?

No, and the reason matters more than the answer. Work the two offers as sums and they come out the same. The gamble averages Rs 5,500/- in the first wording and costs an average of Rs 5,500/- in the second. The certain option is Rs 5,000/- both times. A calculation error would have to show up somewhere, and there is nowhere for it to show up: the sums are two multiplications and nobody in the room got them wrong.

Nor is it a failure of reading. The same 60 people met both wordings, so any suggestion that the loss wording was harder to follow has to explain how the identical people followed the gain wording perfectly well. Nothing numerical differs between the two offers, so the reversal cannot be a calculation error, and it happens before any calculation begins. By the time the arithmetic starts, the frame has already fixed the starting point the arithmetic works from.

The distinction is worth sitting with. Knowing that the arithmetic is sound changes what can be done about the effect. If the effect were an arithmetic error, the response would be to check the arithmetic, and checking the arithmetic here catches nothing at all. The two sums agree. The response has to be aimed upstream, at the description, and that is a different kind of check with a different place in a process.

Is there such a thing as a neutral description?

Everybody asks this question next, and the honest answer is no. A neutral descriptionA statement of an outcome that implies no starting point at all. Useful to imagine and impossible to write. would be a statement of an outcome that implies no starting point whatsoever, and any attempt to write one runs into the fact that every sentence available implies something. Even the barest possible statement, a holding is worth Rs 2,55,000/-, implies that the right way to think about it is as a quantity rather than as a movement. Treating an outcome as a quantity is itself a position, not an absence of one.

Take one ordinary result and write it four ways. The Nilgiri mid-cap scheme in the invented log cost Rs 3,00,000/- and stands at Rs 2,55,000/- on 30 September. Measured against what was paid, it is down Rs 45,000/-, being 15.0 per cent. Measured against the Rs 3,45,000/- somebody had in mind when they bought it, it is short by Rs 90,000/-, being 26.1 per cent. The Vindhya index scheme held beside it is up 12.0 per cent, so measured against that the Nilgiri scheme is 27.0 points behind. Measured against the whole holding, where Rs 13,00,000/- went in, the Rs 45,000/- shortfall is 3.5 per cent of what was committed.

Look at what those four have in common: each is arithmetically correct, each is the kind of sentence that appears in ordinary statements, and each would produce a different reaction in a reader. Four ordinary ways of stating one result are four frames, not one neutral statement and three distorted ones. The sorting cannot be done, and the instinct to sort them into the honest one and the loaded ones is the instinct worth interrupting.

Who chooses the frame in an ordinary sentence?

Whoever writes the sentence. The whole answer is that short, and the interesting part is how rarely anybody notices they are choosing. A sentence about a result has to start somewhere. The writer reaches for the number closest to hand. The nearest number is usually the purchase cost because that is what the record shows, and the frame is set before any thought about framing has occurred to anybody.

The frame is chosen by whoever writes the sentence, and most of the time nobody involved notices that a choice was made at all. An unnoticed choice is what makes framing so common compared with the effects people worry about more. Framing does not need anybody to be trying to influence anybody. It needs no motive, no strategy and no attention. A statement generated by a system that reports every position against its cost has framed every line on it, and the system has no intentions whatsoever.

Deliberate framing exists too, of course. The useful skill here is detection, and detection protects the person doing the reading rather than the person doing the writing. Noticing a frame in something that has been handed over, and then restating it, is the skill worth having.

What does a frame imply about the point an outcome is measured from?

Every frame implies a reference pointThe level an outcome is judged as a movement away from. Rarely stated, and almost always doing work., and the implication is where the detection work happens. A sentence rarely announces its starting point. A sentence just uses a starting point, and the reader absorbs it along with the sentence. So one question pulls a frame into the open, and it is not whether the sentence is fair. Fairness is unanswerable. Ask instead which starting point the sentence measures from, and whether anybody chose that point on purpose.

Try it on one position in the invented log. Kesari Logistics Limited cost Rs 3,00,000/- and stands at Rs 1,95,000/- on 30 September. By 31 March following, it has fallen a further 20.0 per cent to Rs 1,56,000/-, a further Rs 39,000/- gone. Now: is the holding down Rs 1,44,000/-, or down Rs 39,000/-, or worth Rs 1,56,000/-? All three sentences are true on 31 March. The three sentences differ only in which of three starting points they measure from, and each one leaves a person feeling differently about a position that is exactly what it is.

Notice the arithmetic in that picture. Every figure in it is worth checking rather than taking. Rs 3,00,000/- less Rs 1,56,000/- is Rs 1,44,000/-, and Rs 1,44,000/- on Rs 3,00,000/- is 48.0 per cent. Rs 1,95,000/- less Rs 1,56,000/- is Rs 39,000/-, and Rs 39,000/- on Rs 1,95,000/- is 20.0 per cent. Both are correct and they describe the identical holding on the identical day. The quickest way to detect a frame is to ask what point the sentence is measuring from, and then to ask whether anybody chose that point or simply inherited it from whatever the record happened to store.

Why a starting point has such a grip once it has been supplied is set out under prospect theory. The explanation is well worked out and belongs with preferences rather than with wording, so the effect is worth meeting in ordinary choices before the machinery behind it.

What can be done when every description is a frame?

The response is procedural rather than perceptual. Trying to read a description correctly does not work, because there is no correct reading available and because the effect operates on people who know about it. The response that does work is mechanical, boring, and takes about ten minutes: before anything is decided, the same outcome is written out from two other starting points, and the reaction to each is then examined.

The test is what happens next. If the choice that was about to be made changes when the starting point changes, the wording was doing the work and the effect has just been caught. If the choice holds across all three, then whatever is driving it is something other than the sentence, and that is now known rather than merely hoped. Restating an outcome from two other starting points converts an invisible influence into a visible test, and the test takes ten minutes rather than a week.

Two cautions come with the test. The first is that it is worth doing before the decision rather than after it, because afterwards the restatements arrive as a defence of a choice already made. The second is that three wordings is a working number, not a rule: two is enough to catch the obvious cases and four is rarely worth the time. A single wording cannot be compared with anything, so what matters is that more than one exists.

How does somebody work with this, alone or on behalf of others?

The household version comes first. A household needs no adviser and no committee to run it. The invented case has Rs 1,80,000/- of card borrowing running at 36.0 per cent a year. The same cost written three ways shows what each wording does. Written as a rate, it is 36.0 per cent a year and sounds like a line in a statement. Written as a yearly amount, it is Rs 64,800/-, being 36.0 per cent of Rs 1,80,000/-. Written as a monthly amount, it is Rs 5,400/-, one twelfth of Rs 64,800/-. One contract, three sentences, all correct, and most people react hardest to exactly one of the three and barely at all to the other two.

The three wordings of the borrowing cost differ by nothing except the starting point each one implies, so a person who reads only one of them has had their reaction chosen for them by whoever wrote it. The rate invites a comparison against other rates. The yearly amount invites a comparison against a year of something else, a holiday, a fee, a repair. The monthly amount invites a comparison against the rest of the month's outgoings. Three comparisons, three feelings, one debt.

Now the professional version. There is no neutral statement of how a holding has done, so Devika Rao, the adviser at the invented Palash Advisory Services Private Limited, cannot hand Meera Sundaram one. Devika Rao can give the same result from more than one starting point in the same conversation: against what was paid, against where the position stood at the last review, and as a plain quantity. Restating is not a softening technique and it is not a hardening one. Giving several starting points is the only way of putting a client in a position to notice which parts of their own reaction are about the holding and which parts are about the sentence.

The same move works for anyone reading somebody else's numbers. A lender presenting a loan, a scheme document setting out a result, an employer describing a change to a benefit: in each case the reader can restate the same fact from a different starting point before responding to it, and the restating costs nothing.

Meera Sundaram, Devika Rao, Palash Advisory Services Private Limited, the Palash decision log, the Palash 100 index, the Vindhya index scheme, the Nilgiri mid-cap scheme, Suvarna Chemicals Limited and Kesari Logistics Limited are invented.
Educational material. Not advice on any investment, tax, budget or market position.