As anyone who has been part of or through a financial audit knows, red flags are fairly common. Most people in large corporations get dragged into an audit at some point so you know what I’m talking about.
Red flags raised by auditors don’t mean there’s fraud. A few flags can mean a lot of things: sloppy records, random stuff, individual stupidity, plus, of course, outright dishonesty.
For example, there are things like this
Surely, just an innocent mistake. Or, plain stupidity. No doubt, they’ll “fix the glitch.”
But a lot of red flags are often a sign of criminal behavior somewhere.
See one cockroach, you know there are more.
One of the flags auditors look for is inordinate populations in the tail of the curves. Numbers that are too random. Or, not random enough – too good to be true.
Or, perhaps anomalies in the distribution. That’s often a sign of fraudulent behavior.
Sort of like when you see carpenter ants – you know there’s rot somewhere in the structure. Just not obvious where.
But auditors use a range of tools to test for fraud. One such tool is the Newcomb-Benford law.
The Newcomb–Benford law (NBL) defines a probability distribution describing the expected patterns of significant digits in real positive numbers. NBL tells you how the distribution should look like.
And, in real life, that distribution should not be random.
To the contrary, NBL observes the first significant digits are not uniformly scattered. Rather, the digits follow a logarithmic-type distribution.
Auditors of all kinds – tax, financial, trade data – typically employ NBL to assess the likelihood red flags are, indeed, indicative of fraud .
What’s applicable to process data of all kinds is readily applicable to electoral data.
Lacasa and Fernandez-Gracia  discuss Newcomb-Benford to electoral applications. They demonstrate that fraud can be easily inferred from electoral statistics at the macro and micro level.
Kokak et al.  apply these methods to a number of Russian elections over the years and highlight anomalous behavior consistent with fraud (inter alia, unusual voter turnout percentages and material changes) in the skewed results.
All of this brings us to the current election.
Matt Braynard, a GOP political analyst and former Trump Data Chief, believes he can detect voter fraud by comparing absentee ballots and early voters to the Social Security Death Index and the National Change of Address Database.
He even setup a “GoFundMe” account, though the corporate captains seem to be holding him at arms length.
You can read about him here: https://www.zerohedge.com/political/gop-analyst-raises-170k-purchase-data-conduct-deep-dive-voter-fraud
In the meantime, suffice to say, we will be hearing more about voter fraud in the months to come.
BTW, want to see another take on voter fraud? Check out some Russian elections.
See a pattern in the distribution tail?
Getting a deja vu?
Well, let’s see:
- Biden outperforms Senators in swing states, underperforms in VA, NH, RI
- Biden underperforms Hillary Clinton and Barack Obama in MI, PA, GA, and WI
- Biden mail-in dumps have 100% margins
- Down-ballot senators and congressmen get fewer votes than the top-line presidential candidate
- GOP lose ZERO House seats
Does this look like anomalous behavior?
 Cerioli, A., Barabesi, L., Cerasa, A., Menegatti, M., & Perrotta, D. (2019). Newcomb-Benford law and the detection of frauds in international trade. Proceedings of the National Academy of Sciences of the United States of America, 116(1), 106–115.
 Kobak, D., Shpilkin, S., & Pshenichnikov, M. S. (2016). Statistical fingerprints of electoral fraud? Significance (Oxford, England), 13(4), 20–23.
 Lacasa, L., & Fernández-Gracia, J. (2019). Election Forensics: Quantitative methods for electoral fraud detection. Forensic Science International, 294, e19–e22.