Decision Making
Sunk Cost Task
Also called: sunk cost effect task, sunk cost fallacy task, sunk-cost bias, escalation of commitment, Concorde effect
Measure the sunk cost effect: participants choose between going on and switching to a better option in everyday situations, half after points or time were already spent, scored by how much more often they go on.
5-8 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
- 255
- Evidence grade
- D
- Registered clinical studies
- 0
What it measures
Measures the sunk cost effect. Participants choose between going on with something and switching to a better option in 12 everyday situations, half after points or time were already spent on it and half after nothing was. The main score is how much more often people go on when something was already spent.
Constructs
- Sunk cost effect
- Escalation of commitment
- Decisions after prior investment
- Accounting for time and points
Research uses
- Judgment and decision-making research
- Behavioral economics
- Individual differences in decision making
- Within-study group comparison
How the paradigm works
The sunk cost task measures the sunk cost effect, a greater tendency to continue an endeavor once an investment in money, effort or time has been made (Arkes and Blumer, 1985). In their radar-blank plane problem, 85% of the people told that the plane's research was 90% complete chose to pay for the rest, against 17% of those asked about the same last payment with nothing spent before, although what the plane would bring was the same (as quoted by Tan and Yates, 1995). Each of 12 everyday situations written for this task offers a first option and a second option that the text describes as better, at the same or a lower further cost. Six spend time (a line at a fair, a walk, a call on hold, a trip to a shop, a jigsaw puzzle and an online course) and six spend points, like the points on a shop card or in a game (a concert ticket, a phone game, a rowing boat, a series, drawing classes and a day trip). Each situation has two versions that differ in one sentence: in the spent version it says what was already spent on the first option, and in the nothing-spent version it says that nothing was. The participant sees each situation once, six in each version and three of each kind in each version, drawn from the session's seed, and chooses whether to go on with the first option or switch to the second. There is no time limit, and nothing is earned.
More about the paradigm
One practice situation comes first.
The main score is the sunk cost effect: the share of choices going on with the first option in the spent versions minus the share in the nothing-spent versions, from -1 to 1. Positive means going on more often when something was already spent. The share in each version is also given, with the effect within the time situations and within the points situations: Soman (2001) found a weaker effect for time than for money, and in a registered replication with 821 participants Petrov and colleagues (2023) found it for both when versions with and without a sunk cost were compared. Sunk cost situations agree only weakly with one another within a person (Białek and Biesiada, 2025), so one person's effect is noisy and groups are the better comparison. Scores describe choices in these situations; they are research measures and do not diagnose.
Scores
- Sunk Cost Effectscore
- Share of choices going on with the first option in the spent versions minus the share in the nothing-spent versions, from -1 to 1. Needs at least 2 choices in each version.
- Positive means going on more often when something was already spent; 0 means past spending made no difference
- Going On, Something Spent%
- Share of the situations in the spent version where the participant went on with the first option.
- Informational
- Going On, Nothing Spent%
- Share of the situations in the nothing-spent version where the participant went on with the first option, which the text describes as the worse one.
- Low expected
- Going On, All Situations%
- Share of all choices that went on with the first option.
- Informational
- Effect, Time Situationsscore
- The sunk cost effect within the six situations that spend time. Needs at least 2 choices in each version.
- Informational: compare with the points situations
- Effect, Points Situationsscore
- The sunk cost effect within the six situations that spend points. Needs at least 2 choices in each version.
- Informational: compare with the time situations
- Median Decision Timems
- Median time from a situation'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 screen appeared.
- 0 expected
- Choicescount
- Choices scored, out of 12.
- Informational
Use and evidence
255 papers from 2021 to 2025 name this task in their title or abstract, up 33% compared with 2016 to 2020. That ranks it 108 of 495 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 91 |
| 2024 | 44 |
| 2023 | 36 |
| 2022 | 46 |
| 2021 | 38 |
| 2020 | 42 |
| 2019 | 50 |
| 2018 | 40 |
| 2017 | 29 |
| 2016 | 31 |
| 2015 | 35 |
| 2014 | 25 |
| 2013 | 28 |
| 2012 | 18 |
| 2011 | 24 |
| 2010 | 24 |
| 2009 | 13 |
| 2008 | 11 |
| 2007 | 11 |
| 2006 | 7 |
| 2005 | 8 |
| 2004 | 8 |
| 2003 | 4 |
| 2002 | 4 |
| 2001 | 5 |
| 2000 | 2 |
Use grade C
100 to 299 papers from 2021 to 2025.
Evidence grade D
Little published measurement evidence found.
- Papers on test-retest reliability
- 1
- Papers on reliability or validity
- 26
- Registered clinical studies using it as an outcome
- 0
Note on the counts: includes economics 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.
- Situations
- 12 written for this task. Six spend time: a line at a fair, a walk, a call on hold, a trip to a shop, a jigsaw puzzle and an online course. Six spend points: a concert ticket, a phone game, a rowing boat, a series, drawing classes and a day trip. They are shown in the session's language.
- Versions
- In the spent version, one sentence says what was already spent on the first option, such as 40 minutes in a line or 2,000 points for a ticket. In the nothing-spent version, that sentence says nothing was spent. The rest of the situation, and what each option costs from then on, reads the same in both.
- Choice
- Go on with the first option, or switch to the second, which the text describes as better. Click or tap; no time limit. Whether going on is the first button is drawn from the session's seed and kept for the session.
- Order
- Each situation once: 6 in each version, 3 of the time situations and 3 of the points situations, drawn from the session's seed. No one sees both versions of a situation.
- Points
- The points are part of the stories, like the points on a shop card or in a game. Nothing is earned, and no screen mentions money.
- Practice
- One practice situation, a choice between a picnic and a museum, not scored.
- Scores
- The sunk cost effect: the share going on after something was spent minus the share going on after nothing was. Also each share, and the effect within the time situations and within the points situations.
Arkes and Blumer (1985) gave each version of a problem to a different group, as did Petrov and colleagues (2023), whose registered replication of Soman (2001) found the effect for time as well as for money. Here each person meets both versions, in different situations, and the situations spend points or time, never money. Sunk cost situations agree only weakly with one another (Białek and Biesiada, 2025), so one person's score rests on 12 choices and is best compared across groups. The scores support research 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 between going on with a situation's first option and switching to the second. No choice is marked correct, so accuracy is not judged against chance, and choosing one kind of option throughout is accepted. A run with fewer than half its planned choices is invalid and any other shortfall is flagged; going on in every situation where nothing had been spent is flagged. Choices under 200 ms on at least half the items invalidate.
Why these rules
There is no correct option to compare with chance, so accuracy is not judged against chance. The score compares how often people go on when something was already spent with how often they go on when nothing was, so a person who always goes on or always switches has a real result of 0, and repeated answers are not checked. Where nothing was spent the text describes the second option as better, so going on in all of those situations suggests the situations were not read. 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
This test runs with fixed settings.
Need a different version of this test?
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 Sunk Cost Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Arkes HR, Blumer C, 1985). The main score was sunk cost effect (share of choices going on with the first option in the spent versions minus the share in the nothing-spent versions, from -1 to 1).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Arkes, H. R., & Blumer, C. (1985). The psychology of sunk cost. Organizational Behavior and Human Decision Processes, 35(1), 124-140. https://doi.org/10.1016/0749-5978(85)90049-4
- Tan, H.-T., & Yates, J. F. (1995). Sunk cost effects: The influences of instruction and future return estimates. Organizational Behavior and Human Decision Processes, 63(3), 311-319. https://doi.org/10.1006/obhd.1995.1082Arkes and Blumer's problems and results, quoted.
- Soman, D. (2001). The mental accounting of sunk time costs: Why time is not like money. Journal of Behavioral Decision Making, 14(3), 169-185. https://doi.org/10.1002/bdm.370Sunk costs of time against sunk costs of money.
- Petrov, N. B., Chan, Y. K. M., Lau, C. N., Kwok, T. H., Chow, L. C. E., Lo, W. Y., Song, W., & Feldman, G. (2023). Sunk cost effects for time versus money: Replication and extensions Registered Report of Soman (2001). International Review of Social Psychology, 36(1), 17. https://doi.org/10.5334/irsp.883Versions with and without a sunk cost, for time and for money.
- Białek, M., & Biesiada, E. (2025). On the low reliability of sunk cost vignettes. Brain Sciences, 15(8), 808. https://doi.org/10.3390/brainsci15080808How weakly sunk cost situations agree within a person.
Source papers
- Petrov and colleagues (2023), Study 2, a registered replication of Soman (2001): with time already spent, 34.5% went on with the first project against 7.1% with nothing spent; with money already spent, 31.0% against 10.9%. Each person saw one version.Paradigm source, Adults
Questions
How long does the Sunk Cost Task take?
About 5 to 8 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?
Sunk Cost Effect, Going On, Something Spent, Going On, Nothing Spent, Going On, All Situations, Effect, Time Situations, Effect, Points Situations, Median Decision Time, Choices Under 100 ms and Choices. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Sunk Cost Task used in research?
255 papers from 2021 to 2025 name it in their title or abstract, up 33% compared with 2016 to 2020. That ranks it 108 of 495 test families in the ConductCognition test census (use grade C). Source: ConductCognition test census 0.22.0, counted 3 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Sunk Cost Task have?
In the published research, 1 paper reports 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 languages can participants use?
Participants can take this test in English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese, with its words in the same language. 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.