# Argumentum ex culo

 Part of the series on Key articles General logic Bad logic v - t - e
Argumentum ex culo is the act of arguing by way of non sequitur, ad hockery, or bullshit in a desperate attempt to shore up a blatantly untenable argument.

As a general rule, if something stinks about the argument, there's a good chance it's an argumentum ex culo.

## Otherwise known as

Argumentum ex culo is a Latin translation of PIDOOMA, the acronym for Pulled It Directly Out Of My Ass.[1] This is a self-explanatory origin for woo and pseudoscience statistics, estimates, theories, and, fairly often, evidence, too.

## In statistics

A quoted statistic that is a neat multiple of 5 should generally be suspected to be of PIDOOMA provenance, because after all, "88.2% of statistics are made up".[2] Similarly, statistics or results quoted to odd precision are also suspect as they are an example of "spurious rigor". That is, the addition of significant figures or more decimal places are used to make the figure seem more accurate and reliable, when, in fact, such precision is rarely justifiable. An improvement of 82.56% is probably (with a confidence of either 95% or 92.56584%) no real improvement at all.

While useless to identify a single bogus statistic, large sets of numbers found in nature typically conform to Benford's Law,[3] which states that 30% of the time, the leading digit in each statistic should be "1", 17.6% of the time the leading digit should be "2", 12.5% of the time the leading digit should be "3", and so forth in conformance with the following equation:

$P(d)=\log_{10}(d+1)-\log_{10}(d)=\log_{10} \left(1+\frac{1}{d}\right).$

Made-up statistics, unlike naturally occurring statistics, tend to have an even distribution of first digits. Keep in mind this rule only applies to purely quantitative numbers, like "flow rate in gallons per second of the Nile river", and percentages, by virtue of their cap at 100% do not conform to this metric.

Conversely, the last digits tend to have a more even distribution as these represent relatively random noise inherent in any real-world sample. Therefore amongst large sets of numbers, an over-prevalence of any particular number (usually 7, for some reason) may suggest these numbers were typed up by a human rather than generated by a noisy and messy environment such as reality.

## Footnotes

1. Synonyms: colonic autoextraction methodology, toilet fishing.
2. Reeves; Guinness (1997)
3. Wolfram MathWorld - Benford's Law