Sources and measurement
Every number.
A named source.
Numerical provenance for the research briefs and the website, with the group each figure describes and the limits of its interpretation.
Read the base before the percentage
Different groups answer different questions. The stage percentages describe separate respondent groups; they are not a longitudinal funnel. The barrier percentages exclude the “nothing” option. The conditional money figure additionally requires a report of feeling stuck.
Event-data verification status
Completed response records in the two Mirai event cohorts, 9–10 June 2026 UTC, marked valid and with completed results. Self-selected and cross-sectional; repeat participation across days is not ruled out. Small subgroup bases require caution. Exact counts are retained privately.
Repeating a figure in an article and a PDF does not create independent evidence. Both use this same primary aggregate reanalysis.
Why the figures changed
Version 1.5 replaces the earlier summary percentages with an explicitly defined cohort. The earlier values could not be reproduced with the checked cohort and completion filters, and their original frozen extract was not located. This analysis includes completed response records marked valid in the two Mirai cohorts on 9–10 June 2026 UTC. It excludes completions outside that window and all other cohort codes. The stored quality flag is a workflow screen, not independent proof of authenticity. Repeat participation across days is not ruled out; the unit is a response record, not an independently verified unique person.
| Figure | What it describes | Source |
|---|---|---|
| 60%, 20%, 56%, 25% | Feeling stuck within curious, learning, choosing and already-investing response groups, respectively. Small bases; separate groups, not a longitudinal funnel. | U Impact (2026), Mirai event-window aggregate reanalysis, 10 October U Impact primary aggregate reanalysis |
| 26%, 21%, 20%, 20%, 13% | Time, knowledge, money, complexity and fear respectively, among responses naming a barrier. The 21 no-barrier responses are excluded; no barrier answers are missing. | U Impact (2026), Mirai event-window aggregate reanalysis, 10 October U Impact primary aggregate reanalysis |
| 26% | Money among responses that both reported feeling stuck and named a barrier. Overlaps the all-barrier group; the difference is not tested or claimed significant. | U Impact (2026), Mirai event-window aggregate reanalysis, 10 October U Impact primary aggregate reanalysis |
| ~30%, about six weeks | Study-specific model attribution of the observed gender difference in financial literacy to confidence; approximate interval between two Dutch household-survey modules. Not prevalence in U Impact participants. | Bucher-Koenen, Alessie, Lusardi and van Rooij (2025), Fearless Woman, Management Science 71(9), 7414-7430 Published primary research paper |
| 43% | Comfortable or very comfortable with risk among responses reporting feeling stuck with a valid risk answer. This is not the share of risk-comfortable respondents who felt stuck. | U Impact (2026), Mirai event-window aggregate reanalysis, 10 October U Impact primary aggregate reanalysis |
What is disclosed
Public reporting contains percentages only. Exact aggregate totals and subgroup counts are retained in the private author audit. Denominator definitions and filters are published, but this page alone does not permit independent numerical reproduction. These figures are descriptive prompts for investigation, not market prevalence estimates.
What the survey asks
The current short event instrument asks “Where are you with investing right now?”, followed by whether the respondent feels they are moving forward or stuck. Its barrier question asks “What’s your biggest challenge right now?”. Risk comfort is asked separately. Wording is checked against U Impact’s survey-v2 implementation as of 30 September 2026; this is not proof that an earlier participant saw an identical instrument.
Configuration is not validation
The full assessment defines seventeen scoring inputs, including four hypothetical scenarios, and a framework of nine profiles. The current event survey uses twelve questions, including three scenarios and one knowledge check. These are configuration counts checked in the survey-v2 source, not findings about participants or proof of predictive performance. Scenario weights and band thresholds are design choices requiring validation for their intended use.
Other cited website figures
The OECD/INFE 2023 report covers 68,826 adults across 39 countries and economies; see the OECD report. The European Commission’s 2023 survey reports 18% high, 64% medium and 18% low overall financial literacy; see Flash Eurobarometer 525. These use different instruments and populations and must not be combined with U Impact event percentages.
Research and practical proposals
Peer-reviewed studies are identified in each PDF. Their results do not validate U Impact’s profiles or prove that a proposed support format works. The explanations, matrices and test briefs are author-proposed ways to investigate a decision. Each report names its joint authors and their commercial interest.
Challenge the evidence
Which question should we investigate next?
Tell us which figure, interpretation or measurement decision you want to discuss.
Email the U Impact team