Decision Making
Escalation Auction Task
Also called: dollar auction, dollar auction game, Shubik's dollar auction, minimal dollar auction
Measure how far people keep bidding when both bidders pay: participants bid in turn against a simulated bidder for a prize of points, scored by how high they go.
4-7 minUntimed
Free to start: 10 free participants per account, then $0.99 per participant for academic research, or any yearly plan. Try it here first, with no account.
- Papers, 2021 to 2025
- 17
- Evidence grade
- D
- Registered clinical studies
- 0
What it measures
Measures how far people keep bidding once both bidders must pay. In each auction the participant and a simulated bidder bid in turn for a prize of 100 points; when one stops, the other wins, and both pay their highest bid. The main score is how high the participant bids against bidders that never stop.
Constructs
- Escalation of commitment
- Sunk-cost reasoning
- Competitive decision making
- Learning from repeated play
Research uses
- Research on escalation of commitment and competitive bidding
- Behavioral economics research on auctions and conflict
- Within-study comparison of bidding over auctions
How the paradigm works
The Escalation Auction Task measures how far people keep bidding once both bidders must pay. It is the dollar auction of Shubik (1971) under our own name: a prize goes to the highest bidder, and the second-highest bidder also pays their bid and gets nothing. Once both have bid, each raise looks worth it to the bidder who is behind, since a small raise could still win the prize while stopping loses what was bid, so bids can climb past the prize's value. Rational bidders would not bid against each other at all (O'Neill, 1986). In a minimal dollar auction most people, experienced managers among them, bid more than the prize was worth and lost money; experience of the same situation prevented later escalation, while setting oneself a limit beforehand did not (Hafenbrädl and Woike, 2018). In repeated auctions between two bidders, experience also reduced escalation (Morone, Nuzzo and Caferra, 2019). How the other bidder is described matters: the median price rose when people were told the opponent might be a person and fell when they were told it might be an optimising machine (Ferraz and Houf, 2026, a preprint).
More about the paradigm
Here the participant and a simulated bidder, a new one in each auction, bid in turn for a prize of 100 points, each bid 10 points above the last, and each has 200 points to bid with. The participant bids first and can stay out at no cost; when one stops, the other wins the prize, and both pay their highest bid. Only the other bidder can reach 200, so a participant who never stops loses 190 points to a bidder who never stops. Each bidder's turn to stop is set before the session: in every block of 6 auctions one stops at its first turn, two at their second, one at its third and two never stop, in an order drawn from the session's seed, and nothing the participant does changes it. Against these bidders the most points come from bidding up to 50 and stopping when the other bidder reaches 60; bidding past the prize's value earns less than stopping at any bid from 30 to 90, and never stopping loses about 17 points an auction. The standard instructions say each other bidder is a computer program that follows set rules; a neutral wording leaves that line out, and the last screen then says so. Studies usually explain the purpose afterwards. The standard session has 12 auctions in 2 blocks (6 to 24). One check question with feedback comes first, there is no time limit, and the study decides what the points are worth. About 4 to 7 minutes.
The main score is the escalation bid: the mean of the participant's highest bid in the auctions whose bidder never stopped, from 0 (staying out) to 190. Above 100 means bidding past what the prize is worth. The share of those auctions in which the participant bid past the prize, the same mean over the first and the last block of auctions and the change between them, the share of auctions not entered, the share won and the points gained or lost are reported too. A high bid can reflect wanting to win, a guess about a program or the points already spent, so the scores describe bids in a game played for points, never a person's judgment or self-control, and they do not diagnose. People who know the bidder is a program may bid differently from people bidding against a person, so compare people who saw the same instructions.
Scores
- Escalation Bidpoints
- Mean of the participant's highest bid in the auctions whose other bidder never stopped, from 0 (staying out) to 190.
- Above 100 is bidding past the prize's value; read across a group
- Bids Past the Prize%
- Share of the auctions whose other bidder never stopped in which the participant's highest bid was more than the prize of 100 points.
- Informational
- Escalation Bid, First Blockpoints
- The escalation bid over the first block of 6 auctions.
- Informational
- Escalation Bid, Last Blockpoints
- The escalation bid over the last block of 6 auctions; empty with one block.
- Informational
- Change in Escalation Bidpoints
- Escalation bid in the last block minus the first; empty with one block.
- Below 0 means stopping sooner with experience
- Auctions Not Entered%
- Share of auctions the participant stayed out of.
- Informational
- Auctions Won%
- Share of auctions the participant won.
- Informational
- Points Gained or Lostpoints
- Points kept over the auctions minus the 200 points each auction gave to bid with.
- Informational
- Check Correctflag
- 1 when the check question on what a bidder who stops pays was answered correctly, 0 when not.
- 1 supports interpretation
- Median Choice Timems
- Median time from a decision's screen to the choice, over choices at or after 100 ms.
- Informational
- Choices Under 100 mscount
- Choices made less than 100 ms after the decision's screen appeared.
- 0 expected
- Auctionscount
- Auctions finished.
- Fewer than planned is flagged for review
- Planned Auctionscount
- Auctions set for the session.
- Informational
Use and evidence
17 papers from 2021 to 2025 name this task in their title or abstract, about the same as from 2016 to 2020. That ranks it 412 of 495 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 4 |
| 2024 | 5 |
| 2023 | 2 |
| 2022 | 2 |
| 2021 | 4 |
| 2020 | 1 |
| 2019 | 3 |
| 2018 | 4 |
| 2017 | 3 |
| 2016 | 6 |
| 2015 | 5 |
| 2014 | 2 |
| 2013 | 3 |
| 2012 | 6 |
| 2011 | 2 |
| 2010 | 5 |
| 2009 | 4 |
| 2008 | 4 |
| 2007 | 0 |
| 2006 | 1 |
| 2005 | 1 |
| 2004 | 0 |
| 2003 | 1 |
| 2002 | 2 |
| 2001 | 1 |
| 2000 | 0 |
Use grade E
Fewer than 30 papers from 2021 to 2025.
Evidence grade D
Little published measurement evidence found.
- Papers on test-retest reliability
- 0
- Papers on reliability or validity
- 0
- Registered clinical studies using it as an outcome
- 0
Note on the counts: includes game-theory papers that run no task.
How the grades work
Use: A is 1,000 or more papers from 2021 to 2025, B 300 to 999, C 100 to 299, D 30 to 99 and E fewer than 30. Evidence: A means at least 25 papers report its test-retest reliability and at least 25 registered clinical studies use it as an outcome; B, at least 8 test-retest papers, and either 5 registered clinical studies or 150 reliability or validity papers; C, at least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies; D, little published measurement evidence found. The counts measure use. They include every paper that names the task in its title or abstract. Grades describe the published research on this task, not ConductCognition's version of it.
Source: ConductCognition test census 0.22.0, counted 3 November 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Each auction
- A prize of 100 points. The participant bids first, and each bid is 10 points above the one before. When one bidder stops, the other wins the prize, and both pay their highest bid. Each has 200 points to bid with, and nobody can bid more; at the first turn the participant can stay out at no cost.
- Bidder
- A computer program, a new one in each auction. In every block of 6 auctions it stops at its first turn in 1, its second in 2 and its third in 1, and never stops in 2, in an order drawn from the session's seed and saved. Nothing the participant does changes it. The standard instructions say each other bidder is a computer program that follows set rules; a neutral wording leaves that line out, and the last screen then says so.
- Auctions
- 2 blocks of 6 (standard; 1 to 4), each auction followed by its result and the points so far. After each choice its outcome shows for 1,200 ms. Points are hypothetical by default; a setting changes the instruction line to say the study pays.
- Check
- One question with feedback before the first auction: the participant bids 50 and stops when the other bidder bids 60, and keeps 150 of their 200 points.
- Scores
- The mean of the highest bid in the auctions whose bidder never stopped, the share of those in which the bid went past the prize, the same mean in the first and last block and the change between them, auctions not entered, auctions won and the points gained or lost.
Shubik (1971) auctioned a dollar to people bidding against one another. Here the bidder is a computer program, the points are ConductScience's own, and each bidder has a limit, so only the bidder can make the last bid. Against these bidders, bidding past the prize's value earns less than stopping at any bid from 30 to 90. People who know the bidder is a program may bid differently from people bidding against a person (Ferraz and Houf, 2026, a preprint), so compare people who saw the same instructions. Studies usually explain the purpose afterwards. The scores describe bids in a game played for points, never a person's judgment or self-control, and do not diagnose.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: the scored run's choices, one bid or stop per decision. No choice is right or wrong, so accuracy is not judged against chance, and bidding on every turn or stopping at once is accepted. A run short of its planned auctions, a wrong check answer and staying out of every auction are flagged for review. Choices under 200 ms on at least half the decisions invalidate.
Why these rules
There is no correct option to compare with chance, so accuracy is not judged against chance. How far to bid depends on what a person expects of the other bidder and on wanting to win, and most people bid past the prize's value in the minimal dollar auction (Hafenbrädl and Woike, 2018), so escalating to the limit is a real way to play, as is never bidding. An answer under 200 ms is faster than a simple visual reaction, so it cannot reflect the item. Following response-time effort screening for untimed tests (Wise and Kong 2005), the session is invalid only when at least half the answers are that fast.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Blocks | 2 | 1 to 4 |
| Bidder wording | A computer program (standard) | A computer program (standard), Neutral: the other bidder |
| Payment line | Hypothetical points (standard) | Hypothetical points (standard), The study pays for points |
Yearly plans can change settings within these limits. Free and pay-as-you-go studies use the standard settings.
Need a setting this test does not have?
Devices and languages
- Untimed test: runs on desktops, laptops, tablets and phones. Participant devices
- Languages: English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese.
- Available on: Every plan, including the free plan.
- No percentile. Scores and quality flags only. How norms work
- Recruiting online? See the setup guides for Prolific, Sona Systems, CloudResearch Connect and Qualtrics.
Cite and describe
Participants completed the Escalation Auction Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Shubik M, 1971; O'Neill B, 1986; Hafenbrädl S, Woike JK, 2018), with the standard settings (blocks: 2; bidder wording: A computer program; payment line: Hypothetical points). The main score was escalation bid (mean of the participant's highest bid in the auctions whose other bidder never stopped, from 0 (staying out) to 190).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Shubik, M. (1971). The dollar auction game: A paradox in noncooperative behavior and escalation. Journal of Conflict Resolution, 15(1), 109-111. https://doi.org/10.1177/002200277101500111The game: both the highest and the second-highest bidder pay, and only the highest gets the prize.
- O'Neill, B. (1986). International escalation and the dollar auction. Journal of Conflict Resolution, 30(1), 33-50. https://doi.org/10.1177/0022002786030001003With a limit on what each can bid, rational bidders would not bid against each other.
- Hafenbrädl, S., & Woike, J. K. (2018). Competitive escalation and interventions. Journal of Behavioral Decision Making, 31(5), 695-714. https://doi.org/10.1002/bdm.2084Most people bid past the prize's value in a minimal dollar auction, experience of the same situation prevented later escalation, and a computer player for online studies needs no deception.
- Morone, A., Nuzzo, S., & Caferra, R. (2019). The dollar auction game: A laboratory comparison between individuals and groups. Group Decision and Negotiation, 28(1), 79-98. https://doi.org/10.1007/s10726-018-9595-5Repeated auctions between two bidders: experience reduced escalation, and pairs bid closer to the equilibrium than individuals.
- Murnighan, J. K. (2002). A very extreme case of the dollar auction. Journal of Management Education, 26(1), 56-69. https://doi.org/10.1177/105256290202600105The most extreme auction its author had seen, in an executive MBA class, and what it shows about emotion, rationality and strategy under competition.
- Ferraz, V., & Houf, L. (2026). People escalate against a competitor labelled human and hold back against one labelled an optimising machine [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2609.21439A repeated all-pay auction: the median price rose when people were told the opponent might be a person and fell when they were told it might be an optimising machine.
Source papers
- Hafenbrädl S, Woike JK (2018). Competitive escalation and interventions. Journal of Behavioral Decision Making, 31(5):695-714.N 1229, Adults
Questions
How long does the Escalation Auction Task take?
About 4 to 7 minutes, as listed in the test library.
What device does a participant need?
Untimed test: runs on desktops, laptops, tablets and phones. The browser must be Chrome, Edge, Firefox or Safari.
What scores does it produce?
Escalation Bid, Bids Past the Prize, Escalation Bid, First Block, Escalation Bid, Last Block, Change in Escalation Bid, Auctions Not Entered, Auctions Won, Points Gained or Lost, Check Correct, Median Choice Time, Choices Under 100 ms, Auctions and Planned Auctions. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Escalation Auction Task used in research?
17 papers from 2021 to 2025 name it in their title or abstract, about the same as from 2016 to 2020. That ranks it 412 of 495 test families in the ConductCognition test census (use grade E). Source: ConductCognition test census 0.22.0, counted 3 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Escalation Auction Task have?
In the published research, 0 papers report its test-retest reliability and 0 registered clinical studies use it as an outcome (evidence grade D in the ConductCognition test census).
Does it report a percentile?
No. Scores and quality flags only.
Which settings can a researcher change?
Blocks (1 to 4), bidder wording (A computer program (standard), Neutral: the other bidder) and payment line (Hypothetical points (standard), The study pays for points). Settings can be changed on yearly plans; free and pay-as-you-go studies use the standard settings.
Which languages can participants use?
Participants can take this test in English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese. English is on every plan, and the other languages come with yearly plans. Have a native speaker read the machine-translated text before launch.
Is this a clinical test?
No. ConductCognition is for research use only. It runs, scores and exports the test; clinical interpretation is outside the platform.
Every test: the full test list.
Run this test in your next study
Start free with 10 free participants on any test, then use participant credits or a yearly plan.
Related tests