Live Sessions

Behavioural science, run live in the room

Everything else on this site is something to read. This is something to do. Scan a QR code at a real talk, take part in a real behavioural economics demonstration alongside everyone else in the room, and watch the results appear as the room answers. Ten sessions are live now, each one a genuine published paradigm, not a party trick.

Pilot session

Anchoring, Live

Two halves of the room see the same question, seeded with two different numbers. Neither side knows it.

Likely mechanismThe first number you see anchors your guess, even when you know it's arbitrary

This runs the same paradigm Tversky and Kahneman used in their original 1974 anchoring study: a comparative judgment against a planted number, then a free estimate of the real answer. Half the room is asked whether the real figure is more or less than a high number; the other half, the same question against a low number. Nobody is told which anchor they got, or that the room split at all. Then everyone guesses the same real fact, and the two groups' answers get compared live.

Try it: Group A Try it: Group B Session & results screen

Browsing on your own, in one browser? Answers you submit on Group A or B log automatically to that same browser's results screen, no extra step. Across separate phones in a real room, each phone only knows its own answer, so the room comparison happens out loud, the results screen has a spot to type those in too.

How this runs at a live talk

  1. Open the session screen first and print the two QR codes shown there (they're permanent, reuse them for every future talk).
  2. Everyone scans, answers the comparative question, then estimates the real fact privately, on their own phone. The whole thing takes under a minute.
  3. The presenter asks a show of hands, or a handful of people per side to call out their number, and types each one into the session screen live, on the projector, right alongside the real answer and a running chart.
  4. The gap is the lesson. Same room, same real fact, same instructions, two different numbers because of a planted anchor nobody agreed to. The screen also shows what Tversky and Kahneman's own 1974 room found, for comparison, and keeps a dated log of every session run on that device, so results from different talks don't get mixed together.
See the full mechanism, citation, and methodology critique: Anchoring.
Live session

Framing Effect, Live

Two halves of the room see the same real outcome, one worded as a gain, one worded as a loss.

Likely mechanismThe same fact feels different depending on whether it's framed as a gain or a loss

This runs Tversky and Kahneman's original 1981 risky-choice framing paradigm. Half the room picks between two options for a public health programme worded in terms of lives saved; the other half picks between the identical two options worded in terms of lives lost. The numbers never change, only the frame does, and the two groups' choices get compared live.

Try it: Group A Try it: Group B Session & results screen

Browsing on your own, in one browser? Answers you submit on Group A or B log automatically to that same browser's results screen, no extra step. Across separate phones in a real room, each phone only knows its own answer, so the room comparison happens out loud, the results screen has a spot to type those in too.

How this runs at a live talk

  1. Open the session screen first and print the two QR codes shown there (they're permanent, reuse them for every future talk).
  2. Everyone scans and picks one of two options privately, on their own phone. Takes seconds.
  3. The presenter asks a show of hands and types the split into the session screen live, on the projector, alongside a running chart.
  4. The gap is the lesson. Same real outcome, same two options, different choices because of how they were worded. The screen also shows Tversky and Kahneman's own 1981 result for comparison, and keeps a dated log of every session run on that device.
See the full mechanism, citation, and methodology critique: Framing Effect.
Live session

Diversification Heuristic, Live

One group splits $300 across a fictional retirement plan's three funds, two of them stock, one bond. The other group splits the same $300 across a different plan, one stock fund, two bond.

Likely mechanismThe mind substitutes an easier question for a hard one: how many funds are there, so what's an even split across them

This runs Benartzi & Thaler's own 2001 “1/n” naive diversification study design directly, already documented on the Principles page. Group A gets a fictional plan with two stock funds and one bond fund; Group B gets a different fictional plan, one stock fund and two bond funds, same $300, same three-fund format. Each person drags sliders to allocate the full $300, and the room compares the average share that ended up in stock per group.

Try it: Group A Try it: Group B Session & results screen

Browsing on your own, in one browser? Answers you submit on Group A or B log automatically to that same browser's results screen, no extra step. Across separate phones in a real room, each phone only knows its own answer, so the room comparison happens out loud, the results screen has a spot to type those in too.

How this runs at a live talk

  1. Open the session screen first and print the two QR codes shown there (they're permanent, reuse them for every future talk).
  2. Everyone scans and drags sliders to allocate their full $300 across their group's three funds, privately, on their own phone.
  3. The presenter asks a show of hands, or a few people per side to call out what share they put in stock, and types each figure into the session screen live, on the projector, alongside a running average.
  4. The gap is the lesson. Same $300, same three-fund format, same room, a different average share ending up in stock depending only on which funds happened to be on the menu. The screen also shows Benartzi & Thaler's own 2001 finding for comparison, and keeps a dated log of every session run on that device.
See the full mechanism, citation, and methodology critique: Diversification Heuristic.
Live session

Gambler's Fallacy, Live

One group sees a coin that just landed heads five times in a row. The other sees the same coin alternating the whole way. Both predict what's next.

Likely mechanismA run that doesn't look balanced gets treated as a debt the next flip is about to repay, even though the coin has no memory of it

This runs the same pattern behind Tversky & Kahneman's 1971 “law of small numbers” finding, now on this site's own coin-flip report. Group A sees a coin that has just landed heads five times in a row and predicts the sixth flip. Group B sees the identical coin alternating heads and tails the whole way and predicts its sixth flip too. The true odds are 50/50 either way; the room's answers show whether the streak pulled anyone toward “tails is due.”

Try it: Group A Try it: Group B Session & results screen

Browsing on your own, in one browser? Answers you submit on Group A or B log automatically to that same browser's results screen, no extra step. Across separate phones in a real room, each phone only knows its own answer, so the room comparison happens out loud, the results screen has a spot to type those in too.

How this runs at a live talk

  1. Open the session screen first and print the two QR codes shown there (they're permanent, reuse them for every future talk).
  2. Everyone scans and predicts the sixth flip privately, on their own phone. Takes seconds.
  3. The presenter asks a show of hands and types the split into the session screen live, on the projector, alongside a running chart.
  4. The gap is the lesson. The coin has no memory, so both groups' true odds are identical, whatever the room actually predicts. The screen also shows Tversky & Kahneman's own 1971 finding for comparison, and keeps a dated log of every session run on that device.
See the full write-up and citation: The Coin Has No Memory: Gambler's Fallacy.
Live session

Sample Size Swing, Live

One group flips 10 coins, the other flips 100. Both watch how far their own result swings from 50/50.

Likely mechanismHow far a result can swing from the true rate shrinks with the square root of the sample size, not the sample size itself

This runs the same square-root law Howard Wainer documented in “The Most Dangerous Equation,” now with a real coin instead of a ranked list. Group A taps a button and the page flips 10 real coins for them automatically. Group B taps the same button and gets 100 flips. Both groups are flipping the exact same fair coin; the room's results show which sample size swings further from 50/50, purely from arithmetic.

Try it: n = 10 Try it: n = 100 Session & results screen

Browsing on your own, in one browser? Flips you submit on either link log automatically to that same browser's results screen, no extra step. Across separate phones in a real room, each phone only knows its own result, so the room comparison happens out loud, the results screen has a spot to log those in too.

How this runs at a live talk

  1. Open the session screen first and print the two QR codes shown there (they're permanent, reuse them for every future talk).
  2. Everyone scans and taps “Flip”: the page itself runs the real flips and submits the exact percentage of heads automatically, no typing required.
  3. The presenter asks a show of hands, or a few people per side to call out their percentage, and logs each one into the session screen live, on the projector, alongside the live spread.
  4. The gap is the lesson. The 10-flip group should swing much further from 50/50 than the 100-flip group, on the exact same fair coin. The screen also shows Wainer's own two real rankings for comparison, and keeps a dated log of every session run on that device.
See the full write-up and citation: The Coin Has No Memory: Why the Smallest Sample Always Looks the Most Extreme.
Live session

Decoy Effect, Live

One group picks from two magazine plans, the other from the same two plus a decoy that never wins.

Likely mechanismAn easy win against a decoy colours a harder comparison it never actually settled

This is built on Huber, Payne & Puto's 1982 decoy paradigm, the same mechanism behind Dan Ariely's well-known Economist subscription menu. Group A picks between Digital only and Print + Digital. Group B picks between the same two options plus a third, Print only, priced the same as the bundle but giving strictly less. The decoy never wins; it just shifts which of the other two does.

Try it: Group A Try it: Group B Session & results screen

Browsing on your own, in one browser? Answers you submit on Group A or B log automatically to that same browser's results screen, no extra step. Across separate phones in a real room, each phone only knows its own answer, so the room comparison happens out loud, the results screen has a spot to type those in too.

How this runs at a live talk

  1. Open the session screen first and print the two QR codes shown there (they're permanent, reuse them for every future talk).
  2. Everyone scans and picks one option from their group's menu privately, on their own phone.
  3. The presenter asks a show of hands and types the split into the session screen live, on the projector, alongside a running chart.
  4. The gap is the lesson. Adding a losing option shifted the room toward the bundle, without that option ever being chosen. The screen also shows Huber, Payne & Puto's own published range for comparison, and keeps a dated log of every session run on that device.
See the full mechanism, citation, and methodology critique: Decoy Effect.
Live session

Zero Price Effect, Live

One group picks between two cheap chocolates. The other group picks between the same price gap, but the cheap one is free.

Likely mechanismZero removes any chance of a bad trade, so a free option gets picked far more than the same price gap predicts anywhere else on the scale

This runs Shampanier, Mazar & Ariely's 2007 chocolate paradigm. Group A picks between a Hershey's Kiss at 1 cent and a Lindt truffle at 15 cents. Group B picks between the same two chocolates with the identical 14 cent gap between them, except the Kiss is free. Nobody imagines anything: both groups are choosing a real chocolate at a real, stated price, right now.

Try it: Group A Try it: Group B Session & results screen

Browsing on your own, in one browser? Answers you submit on Group A or B log automatically to that same browser's results screen, no extra step. Across separate phones in a real room, each phone only knows its own answer, so the room comparison happens out loud, the results screen has a spot to type those in too.

How this runs at a live talk

  1. Open the session screen first and print the two QR codes shown there (they're permanent, reuse them for every future talk).
  2. Everyone scans and picks one chocolate from their group's choice privately, on their own phone. Takes seconds.
  3. The presenter asks a show of hands and types the split into the session screen live, on the projector, alongside a running chart.
  4. The gap is the lesson. Same price gap, same room, a much bigger swing toward the cheap option the moment it became free. The screen also shows Shampanier, Mazar & Ariely's own 2007 result for comparison, and keeps a dated log of every session run on that device.
See the full mechanism, citation, and methodology critique: Zero Price Effect.
Live session

Present Bias, Live

One group picks a snack a week from now. The other picks the same snack for right now.

Likely mechanismThe value of a reward drops steeply the moment it's delayed at all, so “now” beats “later” out of proportion to the actual wait

This is built on Read & van Leeuwen's 1998 snack-choice paradigm, the same effect Thaler and Benartzi's Save More Tomorrow program was designed around. Group A picks, one week in advance, which snack they'll receive in a week. Group B makes the identical choice for immediate delivery, right now. The options never change, only how far away the reward is at the moment of choosing.

Try it: Group A Try it: Group B Session & results screen

Browsing on your own, in one browser? Answers you submit on Group A or B log automatically to that same browser's results screen, no extra step. Across separate phones in a real room, each phone only knows its own answer, so the room comparison happens out loud, the results screen has a spot to type those in too.

How this runs at a live talk

  1. Open the session screen first and print the two QR codes shown there (they're permanent, reuse them for every future talk).
  2. Everyone scans and picks a snack privately, on their own phone. Takes seconds.
  3. The presenter asks a show of hands and types the split into the session screen live, on the projector, alongside a running chart.
  4. The gap is the lesson. Same two snacks, same room, different choices depending only on whether the reward was a week away or immediate. The screen also shows Read & van Leeuwen's own published result for comparison, and keeps a dated log of every session run on that device.
See the full mechanism, citation, and methodology critique: Present Bias.
Live session

Compromise Effect, Live

One group picks from two cameras, the other from the same two plus a pricier third.

Likely mechanismExtremeness aversion, avoiding the risk of looking wasteful or cheap

This runs Simonson's 1989 compromise paradigm. Group A picks between Camera A and Camera B. Group B picks between the same two cameras plus Camera C, a pricier, more advanced option. Camera B never changes price or features between the two groups, only its position in the lineup does, from top option to middle option.

Try it: Group A Try it: Group B Session & results screen

Browsing on your own, in one browser? Answers you submit on Group A or B log automatically to that same browser's results screen, no extra step. Across separate phones in a real room, each phone only knows its own answer, so the room comparison happens out loud, the results screen has a spot to type those in too.

How this runs at a live talk

  1. Open the session screen first and print the two QR codes shown there (they're permanent, reuse them for every future talk).
  2. Everyone scans and picks one camera from their group's lineup privately, on their own phone.
  3. The presenter asks a show of hands and types the split into the session screen live, on the projector, alongside a running chart.
  4. The gap is the lesson. Adding a pricier third option pushed the room toward the middle camera, the same one that was the top option a moment ago in the other group. The screen also shows Simonson's own published result for comparison, and keeps a dated log of every session run on that device.
See the full mechanism, citation, and methodology critique: Compromise Effect.
Live session

Perceived Fairness, Live

One group judges a price hike after a shortage. The other judges a price hike after a real cost increase.

Likely mechanismA price rise feels fair when it tracks a real cost increase, not just higher demand

This runs two of Kahneman, Knetsch & Thaler's own 1986 fairness vignettes. Group A judges a hardware store that raises the price of a snow shovel from $15 to $20 the morning after a big snowstorm, a price rise that tracks nothing but a spike in demand. Group B judges a grocer who raises the price of lettuce by the exact 30 cents per head their own wholesale cost just rose, after a shortage. Both are real price increases passed straight to the customer; only the story behind them differs.

Try it: Group A Try it: Group B Session & results screen

Browsing on your own, in one browser? Answers you submit on Group A or B log automatically to that same browser's results screen, no extra step. Across separate phones in a real room, each phone only knows its own answer, so the room comparison happens out loud, the results screen has a spot to type those in too.

How this runs at a live talk

  1. Open the session screen first and print the two QR codes shown there (they're permanent, reuse them for every future talk).
  2. Everyone scans and rates their scenario "Fair" or "Unfair" privately, on their own phone.
  3. The presenter asks a show of hands and types the split into the session screen live, on the projector, alongside a running chart.
  4. The gap is the lesson. Two different price rises, judged by two different groups, land on opposite verdicts because one tracks a real cost increase and the other doesn't. The screen also shows Kahneman, Knetsch & Thaler's own 1986 result for comparison, and keeps a dated log of every session run on that device.
See the full mechanism, citation, and methodology critique: Perceived Fairness.
Live session

Biased Assimilation, Live

Two groups get told opposite priors. Both see the exact same result. Then they rate it, and pick the excuse they'd give if someone said they were wrong.

Likely mechanismEvidence that confirms a belief gets accepted; evidence that contradicts it gets picked apart

This runs Lord, Ross & Lepper's 1979 design against five fictional business results instead of one death-penalty study, so you can pick whichever setting lands hardest with your room. Group A is told, before seeing anything, that the team behind the change was confident it would work. Group B is told the team doubted it. Both groups then see the identical result. Everyone rates how convincing it is, then picks which excuse they'd give if someone said they were wrong: something else caused it, it worked for some people, it helped in a different way, or there's a hidden problem. The reveal screen cross-tabs the room's ratings and excuses by which prior they were given, live, so the room watches its own explanations sort themselves by belief, not by evidence. A second view lets anyone re-rate after seeing the split, and compares the room's mood before and after.

Session & results screen

Pick a scenario on the session screen first and print its two QR codes, permanent per scenario, reuse them for every future talk. No QR reader handy? Use the links below instead: for each scenario, send the “Confident” link to half your room and the “Doubtful” link to the other half. Each one opens the tool already set to that version, and the “Copy” button next to it copies the link to paste into a text or Slack message.

A redesigned “what happens next” screen Confident Doubtful
A one-page checkout redesign Confident Doubtful
A redesigned first-run tutorial Confident Doubtful
A redesigned renewal reminder email Confident Doubtful
A weekly meeting cut from 60 to 30 minutes Confident Doubtful

How this runs at a live talk

  1. Pick one scenario on the session screen and print its two QR codes (they're permanent per scenario, reuse them for every future talk).
  2. Everyone scans, gets told a prior, sees the identical result, rates it, then picks an excuse privately, on their own phone. Takes under a minute.
  3. The presenter switches the session screen to the Excuses view to show the room's ratings and excuses cross-tabbed by prior, live, on the projector. Separate phones can call out their rating for manual entry too.
  4. The gap is the lesson. Same identical number, same room, but the confident-primed side rates it more convincing and reaches for a different excuse than the doubt-primed side. Anyone can tap “rate again” once the split is shown; the Revise view compares the room's mood before and after.
See the full mechanism, citation, and methodology critique: Biased Assimilation.
More sessions like this are planned. The same real-research standard (a genuine published paradigm, not a party trick) will apply to whatever comes next.