Reasoning
Growth Prediction Task
Also called: exponential growth bias task, exponential growth estimation task, compound growth estimation task, doubling estimation task
Measure exponential growth bias: participants read neutral scenarios of things that double or grow by a fixed percentage and type how large each will become, scored as the median ratio of their answers to the true amounts.
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
- 31
- Evidence grade
- D
- Registered clinical studies
- 0
What it measures
Measures exponential growth bias. Participants read short scenarios of things that double or grow by a fixed percentage each period, see the first amounts, and type how large each will be after a number of periods. The main score is the median ratio of the answers to the true amounts; below 1 means growth was underestimated.
Constructs
- Exponential growth bias
- Intuitive extrapolation
- Numerical estimation
- Numeracy
Research uses
- Judgment and decision-making research
- Numeracy and financial literacy research
- Risk and science communication research
- Within-study comparison of how growth is described
How the paradigm works
The growth prediction task measures exponential growth bias, the tendency to underestimate how large things become when they grow by a fixed proportion each period (Wagenaar and Sagaria, 1975; Stango and Zinman, 2009). Each of 12 short scenarios, written for this task in each language, states an amount now and its steady growth: six double every period and six grow by a percentage, from 10% to 50%. The amounts now and after 1 and 2 periods are shown, and the participant types the amount after 8 to 25 periods, with no time limit and no calculator. Three scenarios of each kind ask about 8 to 12 periods and three about 15 to 25. The scenarios are neutral things that grow, such as a water plant on a lake, a bamboo grove or the books of a library, with no money, health or illness. Below the answer box the number appears as it will be read, with its digit groups, so the zeros of a long number can be checked. One practice scenario with its answer comes first. A setting runs only the doubling or only the percentage scenarios, to compare the two descriptions between groups.
Scores
- Median Answer Ratioscore
- Median, over the answered scenarios, of each answer divided by the true amount.
- 1 is exact; below 1 means growth was underestimated
- Compounding Weightscore
- Median, over the answers, of the weight alpha that makes (1 + alpha r)^n + (1 - alpha) r n times the start equal the answer (Levy and Tasoff, 2016). Answers at or below linear growth count as 0, and the weight stops at 2.
- 1 is exact compounding; 0 is linear growth
- Underestimates%
- Share of answers below 90% of the true amount.
- Higher means more underestimation
- Close Answers%
- Share of answers within 10% of the true amount.
- Higher is more accurate
- Median Ratio, Doublingscore
- Median answer ratio on the scenarios that double every period.
- 1 is exact; below 1 is underestimation
- Median Ratio, Percentagesscore
- Median answer ratio on the scenarios that grow by a percentage each period.
- 1 is exact; below 1 is underestimation
- Median Ratio, 8 to 12 Periodsscore
- Median answer ratio on the scenarios that ask about 8 to 12 periods.
- 1 is exact; below 1 is underestimation
- Median Ratio, 15 to 25 Periodsscore
- Median answer ratio on the scenarios that ask about 15 to 25 periods.
- Usually lower than over 8 to 12 periods
- Answers Below the Startcount
- Answers smaller than the amount at the start, though every scenario grows.
- Quality check; 3 or more is flagged
- Median Answer Timems
- Median time from a scenario appearing to its answer, for readable answers of at least 100 ms.
- Informational; the task has no time limit
- Answers Under 100 mscount
- Typed answers submitted less than 100 ms after their scenario appeared.
- 0 expected
- Unreadable Answerscount
- Answers left blank or not readable as a number.
- Quality check
- Scenarios Answeredcount
- Answers readable as a number.
- Fewer than half the planned scenarios invalidates
- Scenarioscount
- Scenarios completed in the scored run.
- Informational
- Planned Scenarioscount
- 12 with both kinds of scenario, 6 with one.
- Informational
Use and evidence
31 papers from 2021 to 2025 name this task in their title or abstract, up 72% compared with 2016 to 2020. That ranks it 516 of 670 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 5 |
| 2024 | 10 |
| 2023 | 3 |
| 2022 | 6 |
| 2021 | 7 |
| 2020 | 7 |
| 2019 | 6 |
| 2018 | 1 |
| 2017 | 2 |
| 2016 | 2 |
| 2015 | 3 |
| 2014 | 4 |
| 2013 | 1 |
| 2012 | 5 |
| 2011 | 1 |
| 2010 | 0 |
| 2009 | 1 |
| 2008 | 0 |
| 2007 | 0 |
| 2006 | 0 |
| 2005 | 0 |
| 2004 | 0 |
| 2003 | 1 |
| 2002 | 0 |
| 2001 | 0 |
| 2000 | 0 |
Use grade D
30 to 99 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
Measured on ConductCognition
Pooled across the studies that feed our platform norms, by device group. A figure needs 50 or more people.
No pooled figure yet.
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 0.48.0, counted 12 November 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Scenarios
- 12 short scenarios written for this library in each language: 6 things that double every period and 6 that grow by 10% to 50% a period, such as a water plant on a lake or the books of a library. No money, health or illness.
- Question
- Each scenario states the amount now and its growth, and shows the amount now and after 1 and 2 periods. The question asks for the amount after 8 to 25 periods: 3 scenarios of each kind ask about 8 to 12 periods and 3 about 15 to 25.
- Answer
- A typed number in the answer box of the Numeracy Problems Test, with no time limit and no calculator. Below the box, the number appears as it will be read, with its digit groups.
- Order
- Drawn from the session's seed. One practice scenario with its answer comes first.
- Setting
- Doubling and percentages (standard, 12 scenarios), doubling only or percentages only (6 each), to compare the two descriptions between groups.
- Scores
- Median ratio of the answers to the true amounts (the main score), the compounding weight of Levy and Tasoff (2016), the shares of underestimates and close answers, and the median ratio by kind of scenario and by horizon.
The scenarios and numbers are this library's own; each states its rule and shows exact amounts. The scores describe answers to invented scenarios, support research and do not judge a person's numerical or financial ability.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: growth_prediction_item from the run with the most answered scenarios, latest on ties. Typed amounts have no chance level. Fewer than half the planned scenarios with an answer that reads as a number invalidates. An incomplete run is flagged, and so are 3 or more answers smaller than the starting amount. Answers under 200 ms on at least half of four or more answers invalidate. A hidden tab is flagged.
Why these rules
There is no correct option to compare with chance, so accuracy is not judged against chance. 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. Underestimating growth is the result of interest and never invalidates by itself. At least half the scenarios must be answered for the median ratio to describe the person. Every scenario says the amount grows, so several answers below the starting amount suggest the scenarios were not read; a hidden tab may permit a calculator and is flagged.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| How growth is described | Doubling and percentages (standard) | Doubling and percentages (standard), Doubling only, Percentages only |
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 Growth Prediction Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Wagenaar & Sagaria, 1975; Stango & Zinman, 2009), with the standard settings (how growth is described: Doubling and percentages). The main score was median answer ratio (median, over the answered scenarios, of each answer divided by the true amount).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Wagenaar, W. A., & Sagaria, S. D. (1975). Misperception of exponential growth. Perception & Psychophysics, 18(6), 416-422. https://doi.org/10.3758/BF03204114The first experiment on underestimated exponential growth.
- Stango, V., & Zinman, J. (2009). Exponential growth bias and household finance. The Journal of Finance, 64(6), 2807-2849. https://doi.org/10.1111/j.1540-6261.2009.01518.xExponential growth bias as the tendency to linearize exponential functions, and underestimated future values.
- Levy, M., & Tasoff, J. (2016). Exponential-growth bias and lifecycle consumption. Journal of the European Economic Association, 14(3), 545-583. https://doi.org/10.1111/jeea.12149Typed answers to stated rates and periods, accuracy bands, and the model behind the compounding weight.
- Schonger, M., & Sele, D. (2021). Intuition and exponential growth: Bias and the roles of parameterization and complexity. Mathematische Semesterberichte, 68(2), 221-235. https://doi.org/10.1007/s00591-021-00306-7Growth given as doubling times or as growth rates, the two kinds of scenario.
- Hamann, H. (2022). On getting it right by being wrong: A case study of how flawed research may become self-fulfilling at last. Proceedings of the National Academy of Sciences, 119(15), e2122274119. https://doi.org/10.1073/pnas.2122274119A rereading of the first study, which gave no rule and rounded its numbers; each scenario here states its rule and shows exact amounts.
Source papers
- Levy M, Tasoff J (2016). Exponential-growth bias and lifecycle consumption. Journal of the European Economic Association, 14(3):545-583.N 569, Adults (US panel)
- No normative data for this task.Paradigm source, N/A
Questions
How long does the Growth Prediction 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?
Median Answer Ratio, Compounding Weight, Underestimates, Close Answers, Median Ratio, Doubling, Median Ratio, Percentages, Median Ratio, 8 to 12 Periods, Median Ratio, 15 to 25 Periods, Answers Below the Start, Median Answer Time, Answers Under 100 ms, Unreadable Answers, Scenarios Answered, Scenarios and Planned Scenarios. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Growth Prediction Task used in research?
31 papers from 2021 to 2025 name it in their title or abstract, up 72% compared with 2016 to 2020. That ranks it 516 of 670 test families in the ConductCognition test census (use grade D). Source: ConductCognition test census 0.48.0, counted 12 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Growth Prediction 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?
How growth is described (Doubling and percentages (standard), Doubling only, Percentages only). 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, 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.