How the choice gets framed decides who gets hired

In 1988, Irwin Levin and Gary Gaeth gave people ground beef to evaluate. Half saw it labelled 75% lean, half saw 25% fat. Same beef. The 75% lean group rated it better tasting and less greasy, and their paper in the Journal of Consumer Research reported that the gap shrank once people actually ate it. Experience does push back on wording, but only after the wording has had the first word.

The bigger version came from Amos Tversky and Daniel Kahneman in Science in 1981. They asked people to choose between two programs for a disease expected to kill 600 people. One group read the options as lives saved: save 200 for certain, or take a 1-in-3 chance of saving all 600. 72% took the certain option. A second group read the same two programs as deaths: 400 die for certain, or a 1-in-3 chance that nobody does. Only 22% took the certain option this time. The programs were identical. The verb changed, and a large majority became a small minority.

What I find useful here is that expertise is not the cure. Kahneman spent 40 years on this material and wrote in Thinking, Fast and Slow that his own intuitive judgement was about as prone to these errors as it had been before he started. Framing is a property of how attention works, which is why it runs perfectly well on hiring panels staffed by careful, senior, well-briefed people.

Paul Nutt studied 168 organisational decisions and found that only 29 of the teams looked at more than one alternative. Binary ‘whether or not’ decisions failed 52% of the time over the long run, against 32% for decisions where 2 or more options were really on the table. Chip and Dan Heath called this narrow framing in Decisive, and it’s the thing I point at most often in coaching conversations. The leader has thought hard. All the thinking went into choosing between two options, and none of it into asking where those two came from.

In hiring, how the list is composed does work nobody credits it with. Stefanie Johnson, David Hekman and Elsa Chan looked at 598 job finalists at one university and published the result in Harvard Business Review. With 1 woman in the finalist pool, the odds she was hired were statistically indistinguishable from zero. With 2, the odds of hiring a woman were 79 times higher. With 2 people of colour, the odds of hiring a person of colour were 194 times higher. Nobody in those rooms decided to do that. Whichever group made up the majority of the pool became the default, and the single outlier read as the exception to it.

There’s a fix with evidence behind it. Iris Bohnet, Alexandra van Geen and Max Bazerman found that when evaluators assess candidates side by side rather than one at a time, they weight actual past performance and lean on group stereotypes less. Sequential interviews with a debrief after each one are separate evaluation by design. If your process is built that way, you’ve chosen the frame that makes stereotype cheapest to use. It’s worth reading alongside what we found auditing our own mandate book.

Levin, with Sandra Schneider and Gary Gaeth, later sorted framing into three kinds, and the distinction earns its keep. Risky choice framing moves your appetite for risk, which is the disease problem. Attribute framing moves how good a thing seems, which is the beef. Goal framing moves whether anyone acts at all. The 1998 typology is worth knowing because a hiring process runs all three at different moments, and the fixes aren’t the same.

A role brief is mostly attribute framing, and it’s doing more than describing the job. Danielle Gaucher, Justin Friesen and Aaron Kay went through real job advertisements and found that roles in male-dominated fields used more masculine-coded words: leader, competitive, dominant. In their experiments, published in the Journal of Personality and Social Psychology, women rated the same jobs as less appealing when the wording ran masculine. The mechanism was belonging, rather than any doubt about the skills. They read the ad and concluded the place wasn’t built for them.

Write a mandate as 14 requirements and you’ve described a person. Write it as the two outcomes the role has to produce in its first year and you’ve described a job. Same role, different room turns up, and the second version also gives your panel something to assess against.

Kahneman and Tversky had already named the mechanism in 1979, when they set out prospect theory: an outcome takes its value from how far it sits from wherever we happen to be standing. Their 1981 paper made it concrete with a shopping problem. Buy a jacket for $125 and a calculator for $15, and hear the calculator is $10 at a branch 20 minutes away: 68% said they’d drive. Flip the prices, so the $125 calculator is $120 at the other branch, and 29% would go. Same $5, same drive. What moved was the number people measured the $5 against.

Ron Johnson found out what that costs. He arrived at JCPenney from Apple and in February 2012 replaced the endless coupons and markdowns with plain everyday pricing. On the arithmetic it was a better deal for most shoppers. But what a JCPenney customer had been buying was a $60 shirt marked down to $40, and the markdown was most of the purchase. Take it away and the same shirt at a lower price reads as worse. Revenue fell from roughly $17.3 billion to $13 billion in a year and Johnson was gone in April 2013, 17 months in. Those shoppers were pricing exactly what they had always priced. He’d removed the frame that told them what the number meant.

Every candidate is standing somewhere too. A 15% increase reads as generous against a salary that hasn’t moved in 3 years and reads as an insult against a competing offer they received last week. Same 15%. You can’t address a reference point you haven’t asked about, and most offer conversations never ask.

Simon Gächter and colleagues ran a quiet experiment on academics heading to an economics conference. Everyone got a registration reminder. Some read that late registration carried a penalty; some read that early registration earned a discount. The money at stake was identical either way. Among the PhD students, 93% registered early in the penalty frame and 67% in the discount frame. Of the three kinds of framing, this is the one I’d keep if I could only keep one, because it decides whether anyone moves.

The part of that study people skip is the better part. The senior faculty showed no significant effect. Familiarity with the decision seems to blunt the frame, at least for the bits you’ve priced before. My reading, and it’s a reading rather than their finding, is that this is why senior candidates are hard to move on base pay and surprisingly movable on everything they’ve never had to value: equity in an unlisted centre, the real scope of a mandate, who they’ll actually report to. They have a reference point for the salary. They have none for the rest, so your framing supplies it.

There’s no unframed version of a choice available to you. Presenting a decision one way rather than another is unavoidable, so the only live question is whether the frame you picked tells the truth. 75% lean is true. 25% fat is true. Both describe the same meat, and both facts matter.

The test is simple enough to apply in a meeting. If I’d be comfortable showing the other person the framing I didn’t use, I’m probably fine. If I’d rather they never saw it, I’ve stopped explaining and started selling. A few habits make it workable. Say the choice both ways before you decide, and see whether your preference survives the translation. Swap ‘should we do X’ for ‘what are the ways we could get Y’, and watch the original binary shrink into one option among several. Name the reference point out loud whenever a number appears, including whose reference point it is.

So here’s the question I’d put to you. What’s the last decision you were handed as a yes or a no, that turned out to have more than two answers in it?