Benford's Law Calculator
Paste or type a list of numbers and this tool will extract the leading digit from each value, compare the observed distribution to the theoretical Benford frequencies, and run a chi-square goodness-of-fit test to tell you whether the dataset looks natural or suspiciously manipulated. Results include the chi-square statistic, p-value, mean absolute deviation, per-digit breakdown, and a plain-English fraud-risk interpretation.
What is Benford's Law?
Benford's Law (also called the first-digit law or Newcomb-Benford Law) states that in many real-world datasets, the leading significant digit is not uniformly distributed. Instead, smaller digits appear far more often: the digit 1 leads about 30% of the time, digit 2 about 17.6%, digit 3 about 12.5%, and so on, tapering down to digit 9 at only 4.6%. The relationship is given by P(d) = log10(1 + 1/d), where d is the leading digit. The law holds for data that spans several orders of magnitude and arises naturally from multiplicative processes, such as financial transactions, population figures, river lengths, and physical constants.
How to use this calculator
Paste or type a list of numbers into the dataset field (separate values with commas, spaces, or line breaks). The tool strips signs and leading zeros, extracts the first non-zero digit from each number, and counts how often each digit 1-9 appears. It then runs a chi-square goodness-of-fit test with 8 degrees of freedom to determine whether your observed distribution is statistically consistent with the Benford expectation. You also receive the mean absolute deviation (MAD), a digit-by-digit breakdown table, and a plain-English verdict. Choose the significance level that matches your context: 5% is standard for most analyses, 1% for high-stakes forensic audit use.
Chi-square test and what the results mean
The chi-square statistic summarizes how far each digit's observed count strays from the count predicted by Benford's Law. A higher chi-square means larger deviations. With 8 degrees of freedom, the critical value at the 5% significance level is approximately 15.5: values above this suggest the deviation is unlikely to be random chance. The p-value translates the chi-square into a probability: a p-value of 0.03 means there is only a 3% chance of seeing these digit frequencies if the data were genuinely Benford-distributed. The mean absolute deviation (MAD) offers a more intuitive summary: an average deviation below about 0.006 (0.6% per digit) is considered close conformance; above 0.015 (1.5% per digit) is considered non-conforming in forensic practice.
Fraud detection and limitations of Benford's Law
Benford's Law is widely used in forensic accounting, tax auditing, and election-integrity analysis because fabricated numbers tend to cluster around psychologically comfortable values (e.g., too many 5s and 6s, or an unnaturally uniform distribution). Non-conformance is a red flag that warrants investigation, but it is not proof of fraud. Many legitimately honest datasets also fail the test: assigned numbers (phone numbers, ZIP codes, employee IDs), data constrained to a narrow range (hourly wages, fixed-price items), and small samples all produce non-conforming results for entirely innocent reasons. Always pair this test with domain knowledge and consider second-digit or two-digit analysis for deeper scrutiny.
Benford's Law expected leading-digit frequencies
| Leading digit | Expected probability | Expected % in 1,000 values | Typical MAD threshold |
|---|---|---|---|
| 1 | 30.10% | 301 | < 1.5% acceptable |
| 2 | 17.61% | 176 | < 1.5% acceptable |
| 3 | 12.49% | 125 | < 1.5% acceptable |
| 4 | 9.69% | 97 | < 1.5% acceptable |
| 5 | 7.92% | 79 | < 1.5% acceptable |
| 6 | 6.69% | 67 | < 1.5% acceptable |
| 7 | 5.80% | 58 | < 1.5% acceptable |
| 8 | 5.12% | 51 | < 1.5% acceptable |
| 9 | 4.58% | 46 | < 1.5% acceptable |
Theoretical probabilities P(d) = log10(1 + 1/d) for each leading digit. Natural datasets should match these proportions.
Frequently asked questions
How large does my dataset need to be for Benford's Law to apply?
As a practical minimum, you need at least 50-100 data points for the chi-square test to have reasonable statistical power. With fewer numbers, the test may fail to detect real deviations (low power) or may flag random sampling variation as suspicious. Forensic auditors typically prefer at least 300-500 values. The law works best on datasets spanning three or more orders of magnitude (e.g., transactions ranging from tens to hundreds of thousands).
What types of data naturally follow Benford's Law?
Data generated by multiplicative processes over a wide range typically conforms: financial transaction amounts, invoice values, company revenues, city populations, river lengths, earthquake magnitudes, stock prices, and scientific measurements. Data that does NOT conform includes assigned numbers (phone numbers, SSNs, ZIP codes), data constrained to a narrow range (ages 0-120, hourly wages), and counts that start at zero (e.g., number of items purchased in a single order).
What does a non-conforming result mean?
A p-value below your chosen significance level means the observed digit distribution is unlikely to have arisen by chance from a Benford-distributed population. In forensic contexts this is a flag for further investigation. Common innocent explanations include: the data type is inherently non-Benford (assigned numbers, narrow ranges), the sample is too small, or there is legitimate rounding or truncation in the source system. Only after ruling out these causes should you treat non-conformance as evidence of manipulation.
What is mean absolute deviation (MAD) and what are the thresholds?
MAD is the average absolute difference between each observed digit proportion and its Benford expected proportion, across all nine digits. It is scale-independent, making it easy to compare across datasets of different sizes. Forensic auditing guidelines (Nigrini, 2012) suggest: MAD below 0.006 (0.6% per digit) is close conformance, 0.006-0.012 is acceptable conformance, 0.012-0.015 is marginal conformance, and above 0.015 is non-conformance.
Can Benford's Law detect specific types of fraud?
Different fraud patterns leave different fingerprints. Expense reimbursement fraud (claiming amounts just below an approval threshold, e.g., many values starting with 49 or 99) shows as excess digits 4 or 9. Invented round numbers inflate digits 1-3 or create a spike at specific values. Payroll fraud with fabricated employees often produces too-uniform distributions. While this calculator tests overall first-digit conformance, deeper forensic analysis also examines second digits, two-digit combinations, and the distribution of last digits.
What is the chi-square critical value I should compare against?
With 8 degrees of freedom (9 digit categories minus 1), the chi-square critical values are: 15.51 at the 5% significance level, 20.09 at the 1% level, and 13.36 at the 10% level. If your chi-square statistic exceeds the critical value for your chosen alpha, the p-value falls below alpha and you reject the null hypothesis that the data follows Benford's Law. This calculator computes the exact p-value so you do not need to look up a table.