Statistical uncertainty
The uncertainty from the random fluctuation of event counts, roughly √N for N events.
It shrinks as more data are collected. Many neutrino measurements, for example early T2K appearance or 0νββ searches, are statistics-limited.
In plain words
Neutrino interactions happen at random moments. Even if the true rate is exactly ten per day, one day will bring seven and another thirteen. Statistical uncertainty is the spread that this randomness puts on any measurement based on a finite number of events. It is the one kind of uncertainty that can always be reduced by waiting longer or building bigger.
In detail
- 100 events: about 10% relative uncertainty.
- 10,000 events: about 1%.
- 1,000,000 events: about 0.1%.
For very small counts the Poisson distribution is asymmetric and intervals are computed exactly rather than with √N; for zero events one quotes an upper limit. When background is subtracted, its own fluctuations add to the uncertainty of the signal, so a measurement with N signal events on top of B background events has an uncertainty of roughly √(N + B).
Examples
- In 2011 T2K saw 6 electron-neutrino candidates where about 1.5 were expected without θ₁₃, an indication at about 2.5σ that later grew into a firm observation.
- The roughly two dozen events from SN1987A confirmed the basic picture of core collapse but left energy and timing parameters loosely constrained.
- Daya Bay collected millions of inverse beta decay events, enough that its precision became limited by systematics rather than statistics.
Why it matters
Whether an experiment is statistics-limited or systematics-limited decides what to improve next. If statistics dominate, more exposure helps directly; if systematic uncertainty dominates, more data barely help and the effort shifts to calibration, near detectors and better models.
Frequently asked questions
Why does the uncertainty grow as √N and not N?
Because random fluctuations partly cancel. Doubling the data doubles the expected count but increases the typical deviation only by a factor of about 1.4.
Can statistical uncertainty ever be eliminated?
Not completely, but it can be made arbitrarily small with enough data. Systematic uncertainty, by contrast, stays even with infinite data.
Related terms
Mentioned in 6 pages
Last updated: · Editorial standard
Cite this page
Text
Neutrino Atlas (2026). “Statistical uncertainty”. Retrieved 22 November 2026, from https://neutrino-atlas.com/terms/statistical-uncertainty/BibTeX
@misc{neutrinoatlas2026msstatisticaluncertainty,
title = {Statistical uncertainty},
author = {{Neutrino Atlas}},
year = {2026},
url = {https://neutrino-atlas.com/terms/statistical-uncertainty/},
note = {Accessed 2026-11-22},
language = {english}
}