Social Cognition
Minimum Effort Game
Also called: weak-link game, weakest-link game, minimum game, minimum effort coordination game
Measure coordination when the lowest choice sets everyone's points: over rounds with simulated members, the participant chooses a number from 1 to 7, scored by the mean number and how often it was the group's lowest.
3-6 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
- 24
- Evidence grade
- D
- Registered clinical studies
- 0
What it measures
Measures coordination when the lowest choice sets everyone's points. In each round the participant and simulated members each choose a number from 1 to 7; points rise with the group's lowest number and fall with one's own. The main score is the mean number chosen over the rounds.
Constructs
- Coordination under strategic uncertainty
- Expectations about the weakest member
- Learning from a group's outcome
- Weighing a shared gain against one's own cost
Research uses
- Behavioral economics research on coordination failure
- Studies of teamwork and organizations
- Research on strategic uncertainty and learning in groups
- Group comparison within a study
How the paradigm works
The Minimum Effort Game measures coordination when the lowest choice in a group sets everyone's points. Each member chooses an effort level, here a number from 1 to 7, at the same time. Each member's points rise with the lowest number anyone in the group chose and fall with their own number. Any number chosen by the whole group is an equilibrium, and all choosing the highest pays everyone most, but a high number pays only if everyone else chooses high too, so one low choice holds the whole group back. In the first experiments, in groups of 14 to 16 students, 33 of 107 chose the highest number in the first round and no group's lowest number was above 4; by the tenth round 77 of 107 chose the lowest number, and it was every group's lowest (Van Huyck, Battalio and Beil, 1990). People above the lowest number lowered theirs by less than the gap, and pairs who kept the same partner reached the highest number in 12 of 14 cases. Crawford (1991) explained the fall with an adaptive model: each player moves toward the lowest number, and small random departures below it by any of many players keep pulling it down in large groups. A higher cost of effort lowers efforts, which equilibrium theory does not predict (Goeree and Holt, 2005).
More about the paradigm
Here the participant plays 10 rounds (5 to 30) in a group of 4 (2 to 16), the participant included, with the same simulated members. A member's points are 30, plus 15 times the lowest number in the group, minus 5 times their own number: all choosing 7 gives everyone 100 points and all choosing 1 gives 40, while choosing 7 when the lowest number is 1 gives 10. A table of every case stays on screen. The points are ConductScience's own, with a cost of one third of the gain where Van Huyck and colleagues' was one half. In round 1 each member's number is drawn from the first-round choices of Van Huyck and colleagues' groups. After that, each member moves halfway from its last number toward the last round's lowest number, the participant's included, then one number lower or higher with a 10% chance each, from numbers drawn before the session. So the members never try to raise the lowest number, and it tends to fall over the rounds, as in large groups of people. After each round the participant sees only the lowest number and their points, as in the source. The screens say number, never effort. The standard instructions say the other members are computer programs that follow set rules; a neutral wording leaves that line out, and the last screen then says so. Studies usually explain the purpose afterwards. One check question with feedback comes first, there is no time limit, and the study decides what the points are worth.
The main score is the mean number the participant chose over the rounds. The first number, chosen before any feedback, reflects what a person expects of the group, and the last number and the change from the first half of the rounds to the second show how they followed it. The floor rate is the share of rounds in which the participant's number was the group's lowest, alone or tied. Also reported are the share in which it was below every member's, the share of rounds in which the participant chose the last round's lowest number, the best reply if the others repeat it, and the mean amount by which their number exceeded the lowest, which cost points and raised nobody's. A high number can reflect trust, optimism or an attempt to lead; a low one caution or a guess about a program. So the scores describe choices in a game played for points, never a person's effort at work, and they do not diagnose. People who know the members are programs may choose differently from people in a group of people, so compare people who saw the same instructions.
Scores
- Mean Numberscore
- Mean of the numbers from 1 to 7 the participant chose over the rounds, the effort level.
- Read across a group; no better direction
- First Numberscore
- The number chosen in round 1, before any feedback.
- Informational
- Last Numberscore
- The number chosen in the last round.
- Informational
- Change in Numberscore
- Mean number in the second half of the rounds minus the first half.
- Below 0 means lower numbers with experience
- Floor Rate%
- Share of rounds in which the participant's number was the lowest in the group, alone or tied.
- Informational
- Lowest Alone%
- Share of rounds in which the participant's number was below every member's.
- Informational
- Matched Last Lowest%
- From round 2, the share of rounds in which the participant chose the lowest number of the round before.
- Informational
- Mean Above the Lowestscore
- Mean of the participant's number minus the group's lowest number: numbers that cost points and raised nobody's.
- Informational
- Final Lowest Numberscore
- The group's lowest number in the last round.
- Informational
- Points Wonpoints
- Points over all rounds.
- Informational
- Check Correctflag
- 1 when the check question on reading the table of points was answered correctly, 0 when not.
- 1 supports interpretation
- Median Choice Timems
- Median time from a round'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 round's screen appeared.
- 0 expected
- Roundscount
- Rounds played.
- Informational
Use and evidence
24 papers from 2021 to 2025 name this task in their title or abstract, down 37% compared with 2016 to 2020. That ranks it 386 of 495 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 6 |
| 2024 | 1 |
| 2023 | 10 |
| 2022 | 3 |
| 2021 | 4 |
| 2020 | 7 |
| 2019 | 3 |
| 2018 | 7 |
| 2017 | 13 |
| 2016 | 8 |
| 2015 | 7 |
| 2014 | 3 |
| 2013 | 3 |
| 2012 | 4 |
| 2011 | 7 |
| 2010 | 4 |
| 2009 | 4 |
| 2008 | 3 |
| 2007 | 1 |
| 2006 | 1 |
| 2005 | 4 |
| 2004 | 2 |
| 2003 | 0 |
| 2002 | 1 |
| 2001 | 0 |
| 2000 | 2 |
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
- 1
- 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 round
- Every member of the group chooses a whole number from 1 to 7 at the same time. A member's points are 30, plus 15 times the lowest number in the group, minus 5 times their own number.
- Points
- All choosing 7 gives every member 100 points and all choosing 1 gives 40; choosing 7 when the lowest number is 1 gives 10. A table of every case stays on screen. Points are hypothetical by default; a setting changes the instruction line to say the study pays.
- Group
- 4 members (standard; 2 to 16), the participant included, the same in every round.
- Members
- Computer programs following a rule set before the session. In round 1 each member's number is drawn from the first-round choices in Van Huyck and colleagues' groups: 2, 5, 5, 18, 34, 10 and 33 of 107 for the numbers 1 to 7. From round 2 each moves halfway from its last number toward the last round's lowest number, rounding down, then one number lower or higher with a 10% chance each. Their numbers are drawn from the session's seed before the first round and saved. The standard instructions say the members are computer programs that follow set rules; a neutral wording leaves that line out, and the last screen then says so.
- Rounds
- 10 (standard; 5 to 30), shown as "Round 3 of 10". After each choice the lowest number and the participant's points show for 1.5 seconds, and the next round repeats them.
- Check
- One question with feedback before the first round: the points for choosing 5 when the lowest number is 3.
- Scores
- The mean number; the first and last numbers and the change from the first half of the rounds to the second; the share of rounds at the group's lowest number, alone or tied, below every member's and matching the last round's lowest; the mean amount above the lowest.
Van Huyck, Battalio and Beil (1990) ran groups of 14 to 16 students paid for their points, with a cost of effort half its gain. Here the other members are computer programs, the points are ConductScience's own, with a cost of one third, and the screens say number, never effort. The members move toward the lowest number and never try to raise it, so it tends to fall over the rounds, as in the source's large groups; pairs of people who play together often reach the highest number, and this rule does not. People who know the members are programs may choose differently from people in a group of people, so compare people who saw the same instructions. Studies usually explain the purpose afterwards. The scores describe choices in a game played for points, never a person's effort at work, 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 number from 1 to 7 per round. No number is right or wrong, so accuracy is not judged against chance, and the same number in every round is accepted. A run short of its planned rounds and a wrong check answer are flagged for review. Choices under 200 ms on at least half the rounds invalidate.
Why these rules
There is no correct option to compare with chance, so accuracy is not judged against chance. Every number chosen by the whole group is an equilibrium (Van Huyck, Battalio and Beil, 1990), and 7 or 1 is a sensible choice for someone sure of the others or of nothing, so the same number in every round is a real way to play. 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 |
|---|---|---|
| Rounds | 10 | 5 to 30 |
| Group size | 4 | 2 to 16 |
| Member wording | Computer programs (standard) | Computer programs (standard), Neutral: the other members |
| 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 Minimum Effort Game in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Van Huyck JB, Battalio RC, Beil RO, 1990), with the standard settings (rounds: 10; group size: 4; member wording: Computer programs; payment line: Hypothetical points). The main score was mean number (mean of the numbers from 1 to 7 the participant chose over the rounds, the effort level).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Van Huyck, J. B., Battalio, R. C., & Beil, R. O. (1990). Tacit coordination games, strategic uncertainty, and coordination failure. American Economic Review, 80(1), 234-248.
- Crawford, V. P. (1991). An "evolutionary" interpretation of Van Huyck, Battalio, and Beil's experimental results on coordination. Games and Economic Behavior, 3(1), 25-59. https://doi.org/10.1016/0899-8256(91)90004-XAn adaptive model of the fall: players move toward the lowest effort, with small random departures.
- Goeree, J. K., & Holt, C. A. (2005). An experimental study of costly coordination. Games and Economic Behavior, 51(2), 349-364. https://doi.org/10.1016/j.geb.2004.08.006A higher cost of effort lowers efforts.
Source papers
- Van Huyck, Battalio and Beil (1990): in groups of 14 to 16 students paid for their points, 31% chose the highest number in the first round and 72% chose the lowest by the tenth, when every group's lowest number was the lowest possible. Other points, partners and group sizes will give other numbers.Paradigm source, Adults
Questions
How long does the Minimum Effort Game take?
About 3 to 6 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?
Mean Number, First Number, Last Number, Change in Number, Floor Rate, Lowest Alone, Matched Last Lowest, Mean Above the Lowest, Final Lowest Number, Points Won, Check Correct, Median Choice Time, Choices Under 100 ms and Rounds. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Minimum Effort Game used in research?
24 papers from 2021 to 2025 name it in their title or abstract, down 37% compared with 2016 to 2020. That ranks it 386 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 Minimum Effort Game 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?
Rounds (5 to 30), group size (2 to 16), member wording (Computer programs (standard), Neutral: the other members) 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.