The Odds in a Word
See how much chance people hear in words like “likely” and “might happen.”
Try the project ↓How to use this project
Help
How to explore
Forecasts, warnings, and everyday plans often use words instead of numbers. This project shows how differently people can read the same phrase.
Set your estimate
Choose a phrase and set the chance you think it describes before revealing the group response.
Compare all phrases
See the middle answer and the middle half of answers for each of the 19 phrases.
Try a pair
Put two phrases together and see which one participants usually read as the higher chance.
Check the limits
Use the results as evidence of disagreement, not as a dictionary for probability words.
- Probability
- A chance written from 0%, meaning no chance, to 100%, meaning certain.
- Median
- The middle answer after all numerical estimates are put in order.
- Middle half
- The range from the 25th to the 75th percentile, containing the middle 50% of answers.
- Direct comparison
- A quiz choice between two phrases, asking which one describes the higher chance.
19 phrases · 5,174 people
The odds
in a word
See how much chance people hear in words like “likely” and “might happen.”
Forecasts, warnings, and everyday plans often use words instead of numbers. The same word can leave two people with different expectations.
Start here
What does “likely” mean to you?
Set the chance you hear in this phrase. Then compare your answer with all 5,174 quiz participants.
Move the slider, then reveal the full response pattern.
01 / Compare all phrases
How did people translate each phrase into a number?
Each line runs from 0% to 100%. The thick band holds the middle half of answers, and the dot marks the median. Choose a row to use that phrase in the first view.
“About Even” centers tightly on 50%. “Realistic Possibility” covers the widest middle range: 40% to 75%. These are descriptions of this self-selected quiz group, not fixed meanings for the words.
Which phrases stretched the most?
The bar shows the width of the middle half of answers. A longer bar means participants gave a wider range of numbers to the same phrase.
Read the values for all 19 phrases
| Phrase | Median | Middle half | Mean |
|---|---|---|---|
| Almost No Chance | 2% | 1%–5% | 3.9% |
| Remote Chance | 5% | 3%–10% | 8.7% |
| Highly Unlikely | 5% | 5%–10% | 9.2% |
| Little Chance | 10% | 5%–15% | 11.7% |
| Chances are Slight | 10% | 5%–15% | 13% |
| Improbable | 10% | 5%–20% | 13.4% |
| Unlikely | 20% | 10%–25% | 19.0% |
| Could Happen | 40% | 25%–50% | 39.7% |
| Might Happen | 40% | 25%–50% | 40.0% |
| May Happen | 40% | 30%–51% | 41.9% |
| About Even | 50% | 50%–50% | 50.0% |
| Realistic Possibility | 60% | 40%–75% | 57.0% |
| Better than Even | 60% | 55%–60% | 58.1% |
| Probable | 75% | 65%–80% | 71.4% |
| Likely | 75% | 67%–80% | 72.6% |
| Very Good Chance | 80% | 75%–85% | 79.2% |
| Highly Likely | 90% | 80%–90% | 85.3% |
| Almost Certain | 95% | 90%–98% | 94.0% |
| Will Happen | 100% | 99%–100% | 97.6% |
02 / Put two phrases head to head
Which phrase sounds more likely?
Choose two phrases. The split shows how often participants picked each one when the two appeared together. The median numbers below come from a separate part of the same quiz.
357 direct comparisons · numerical medians: Likely 75%, Probable 75%
The group was nearly split. These phrases did not establish a clear order. A direct choice can separate words that share the same median, but it still does not create a universal definition.
Why this exists
The question
How consistently did 5,174 quiz participants translate 19 everyday probability phrases into numbers, and which phrases were hardest to order?
What to try
Set the chance you hear in one phrase, compare it with the group, scan all 19 phrases on the same scale, then put any two phrases head to head.
Method and limits
About the data
I discovered Adam Kucharski's CAPphrase dataset through TidyTuesday's March 10, 2026 brief and retrieved the underlying source on September 23, 2026. The source is licensed CC BY 4.0 and pinned to commit f12d35545011d83cee3803ebf21be675c9d254da. The checked-in Python script verifies SHA-256 hashes b56b7b380ae5d8743cfdddee0fd0ec08370940d163a574d916d3f555a68a3dfb, 32e0387bc3f39a66ceef9b3e22213a3c30597df4b000af2505dd2cf9ad27d81a, and af44b73e2a18501b96ecd2babc12a019be97e33a838e4f87866282a2cf7fc835 for the three source files. Those files contain 5,174 unique participant records, 98,306 numerical estimates across 19 phrases, and 51,740 pair-choice rows. The static 23,327-byte reduction keeps counts, means, medians, percentile ranges, histograms, and aggregate pair totals; it contains no respondent records. The head-to-head view uses only pair positions 1 through 9, leaving out the tenth choice because it repeated the first pair to test consistency. That leaves 46,566 choices across all 171 possible phrase pairs. Percentiles use linear interpolation between ordered observations. Participants opted into an online quiz and optional background questions, so the group is not a representative sample. Context, language background, and the stakes of a decision can affect interpretation. The project describes the recorded responses but cannot establish correct phrase meanings or explain why people disagreed.
Skills used
- Data analysis
- Data engineering
- Statistical reasoning
- Information design
- Interaction design
- Visual design
- Editorial storytelling
- Accessibility
- Performance
- Product thinking
What the data shows
- “Likely” and “Probable” both had a median estimate of 75%. In their 357 first-pass direct comparisons, 53.5% chose “Likely” as the higher chance and 46.5% chose “Probable,” a near-even split.
- “Realistic Possibility” had the widest middle half of numerical answers, from 40% to 75%. “About Even” had the narrowest: both its 25th and 75th percentiles were 50%.
- “Could Happen” and “Might Happen” both had a median estimate of 40%. Their 224 first-pass direct comparisons were also nearly split: 51.3% chose “Could Happen” as higher.