How accurate is FRQuick?
On 108 human-graded AP History essays, FRQuick's rubric score landed within one point of the human score 92.6% of the time, with an average quadratic weighted kappa (QWK) of 0.88 and a mean absolute error of 0.62 points. 8 of the 108 essays were off by two or more points.
Results by essay type (2026)
Our headline stats across 108 human-graded AP History essays: average QWK 0.88, 92.6% within ±1 point, and mean absolute error 0.62 points. QWK measures how closely two graders agree, adjusting for chance. Learnosity's guide classifies 0.80+ as almost perfect agreement and 0.61–0.80 as substantial agreement.
AP History Long Essay (LEQ)
- Essays
- 60
- QWK
- 0.88
- Within ±1 pt
- 93.3%
- Avg error
- 0.63 pt
4 essays off by 2+ points
AP History Document-Based Question (DBQ)
- Essays
- 48
- QWK
- 0.89
- Within ±1 pt
- 91.7%
- Avg error
- 0.60 pt
4 essays off by 2+ points
How was the benchmark measured?
Each essay already carried a human score. FRQuick graded it again with the same published AP rubric and the same grader students use, and we compared the two total scores: how often they landed within one point, how far apart they were on average, and how well they agreed after accounting for chance (QWK).
What doesn't the benchmark cover?
The published numbers cover AP History DBQ and LEQ essays only. AP Lang and AP Lit essays are graded with their own published rubrics, but we don't publish accuracy figures for them yet; they'll be added once they're measured on essays the grader has never seen.
These are practice-calibration checks on our internal dev set, not independent validation or official College Board certification. Sample sizes vary by essay type, and we only publish numbers for rubrics we have benchmarked thoroughly. Other question types are graded with the same rubrics but are still being evaluated.
How should you use a FRQuick score?
As a practice estimate. A score within a point of a trained reader is close enough to tell you which rubric row to work on next, but any single score can still miss, and your real AP exam is always scored by human readers. See how scoring works, or try it on your own essay with the DBQ grader or LEQ grader.