How it works
For natural data spanning multiple orders of magnitude, the expected frequency of leading digit d is:
P(d) = log₁₀(1 + 1/d)
The pattern was first observed by astronomer Simon Newcomb in 1881 (the first pages of logarithm tables wear out faster) and validated in 1938 by physicist Frank Benford on more than 20,000 real-world numbers. Its reach is astonishing: river lengths, stock volumes, populations, physical constants — and real experimental measurements.
Why it catches fabricated data
When people are asked to invent “random numbers”, nearly everyone makes 1-9 appear roughly equally often — because intuitively “random = uniform”. But the true natural distribution is strongly skewed. Both kinds of deviation are informative:
- Leading digits too uniform — a hallmark of invented “random” data
- Leading digits too concentrated (say, all 1-3) — data may have been adjusted or truncated by hand
The test: a chi-square goodness-of-fit test on the leading-digit distribution (df = 8); p < 0.05 indicates a significant deviation from Benford's expectation.
Key applicability limits (false-alarm sources)
- Data must span multiple orders of magnitude (at least 2). Data confined to 100-200 is out of scope for Benford — this tool warns automatically.
- Bounded-range data doesn't apply: percentages (0-100), scale scores (1-7), ratios, etc.
- Sample size < 100 gives low expected frequencies; interpret p-values with caution — the tool warns.
- Data rounded to few digits (e.g., always two decimals) can introduce slight bias.
Classic applications
Benford's Law is an established tool in financial fraud screening(used by accounting firms and tax authorities for decades), and in research data forensics it is combined with terminal digit analysis, GRIM, and other methods. A single Benford deviation is never proof of fabrication — it is a lead worth questioning.
Sources
Newcomb S (1881). Note on the frequency of use of the different digits in natural numbers. American Journal of Mathematics, 4(1), 39–40.
Benford F (1938). The law of anomalous numbers. Proceedings of the American Philosophical Society, 78(4), 551–572.
Nigrini MJ (1996). A taxpayer compliance application of Benford's law. Journal of the American Taxation Association, 18(1), 72–91.
⚠️ Important: Benford's Law and every method on this site are screening aids. No single anomaly is sufficient to determine data fabrication. Final determinations must be made by competent institutions through due process.