# Benford's Law and Election-Fraud Claims ## The real result Benford's law describes the distribution of leading digits in many naturally occurring numerical data sets. The first digit is 1 about 30.1% of the time, 2 about 17.6%, declining to 9 at about 4.6%. The general form is: > P(first digit = *d*) = log₁₀(1 + 1/*d*) The law holds for data spanning several orders of magnitude and generated by multiplicative or growth-like processes — river lengths, city populations, physical constants, stock prices, and, importantly, many kinds of financial data. It is a genuine and useful tool. Forensic accountants use it to flag books for further examination, and the IRS has employed it for exactly that purpose. **It is a screening heuristic, not a test of fraud.** A Benford deviation says "look here," never "fraud occurred." ## Why it does not apply to precinct vote counts The law's precondition is that the data span multiple orders of magnitude. Precinct vote totals do not. American precincts are deliberately sized to a comparable number of voters — typically a few hundred to a few thousand. A candidate's precinct totals therefore cluster in a narrow band. If most precincts deliver a candidate between 200 and 900 votes, the leading digits will be distributed by the shape of that particular band, not by Benford's logarithmic curve, and the deviation from Benford is guaranteed *before any votes are counted*. The failure mode is sharper still for two-candidate races. When two candidates split a bounded total, their leading-digit distributions are mechanically coupled — one going up forces the other down. Benford's law makes no prediction about such data and never claimed to. **This is not a subtle point and it is not disputed among statisticians.** The American Statistical Association and numerous election-forensics researchers addressed it directly in 2020. The consensus is not that Benford analysis showed no fraud; it is that Benford analysis is the wrong instrument for this data and its output carries no information about fraud either way. ## How it was used Following the 2020 election, Benford charts of precinct returns circulated widely as proof of manipulation, presented in [[Stop the Steal]] material and amplified across the [[Election Denialism]] ecosystem. The charts were real plots of real data. The analysis was inapplicable. A revealing detail: **the same analysis applied to the other candidate's totals in the same precincts frequently showed comparable or larger deviations.** Presenters generally displayed only one. This is the signature of motivated selection rather than of a methodological misunderstanding — a genuine mistake would not consistently omit the control. ## Why this belongs in the branch Benford abuse is the model instance of **counterfeit rigor**, and it is more instructive than ordinary misinformation for three reasons: **The underlying mathematics is real.** There is no false claim to fact-check. Benford's law exists, works, and is used professionally. The deception is entirely in the application. **It is unfalsifiable to the target audience.** Rebutting it requires explaining scale invariance and the order-of-magnitude precondition. That is a harder communicative task than the accusation, and the asymmetry is the whole point — this is [[Manufacturing Doubt]] operating through technical asymmetry rather than through competing claims. **It confers borrowed authority.** A logarithmic plot with a fitted curve looks like analysis. The visual grammar of statistics does the persuasive work that the statistics cannot. The general form: **importing a valid technique into a domain where its assumptions fail, and relying on the audience's inability to check the assumptions.** See [[Statistical Impossibility Arguments]] and [[Curve-Fitting and Cherry-Picked Windows]] for the sibling patterns, and [[Numerology]] and [[Gematria]] for the versions that dispense with the underlying mathematics entirely. ## Related [[Statistical Impossibility Arguments]] · [[Curve-Fitting and Cherry-Picked Windows]] · [[Pseudo-Precision as Rhetoric]] · [[Gematria]] · [[Numerology]] · [[Election Denialism]] · [[Stop the Steal]] · [[2000 Mules]] · [[Manufacturing Doubt]] · [[Mathematics and Power Hub]]