Verified Data Analysis
AI analysis is a draft. Here you'll verify an AI-generated summary against the data — recompute a number, check a chart claim, and catch the result that isn't real.
Work through it
The data & the AI summary
Data (de-identified): a CSV of quiz scores. Week 1 quiz is graded out of 50; Week 5 quiz is graded out of 100. Raw class averages: Week 1 = 36/50, Week 5 = 80/100.
AI summary says: (1) "Average score nearly doubled — up from 36 to 80." (2) "A bar chart shows a steady upward trend." (3) "Participation correlated strongly with score (r = 0.81)."
Your verification
Check each claim. Saved on this device.
| AI claim | Your check | Holds? (yes/no) |
|---|---|---|
| "Nearly doubled (36→80)" | ||
| "Steady upward trend" chart | ||
| "r = 0.81" correlation |
Saved automatically on this device.
Reveal model answer
| Claim | Check | Verdict |
|---|---|---|
| "Nearly doubled" | Put both on a common scale: 36/50 = 72%, 80/100 = 80%. The real gain is ~8 points (about +11%) — not a doubling. | NO — the AI compared raw scores across two different maximums (out of 50 vs out of 100). |
| "Steady upward trend" | Only two data points exist — you can’t claim a "steady trend" from two quizzes. | NO — overstated; the chart implies more than the data supports. |
| "r = 0.81" | No participation column was provided. The correlation can’t have been computed from this data. | NO — hallucinated result; ask to see the code/columns. |
Defensible summary: "On a common percentage scale, the class average rose from 72% (Week 1) to 80% (Week 5) — about an 8-point gain, not a doubling. With only two quizzes, no trend or correlation can be claimed; the "nearly doubled" jump and r = 0.81 are not supported by the data."