Testing Benford’s Law

Imagine a large dataset, say something like a list of every country and its population.

Country	Population
Afghanistan	29,117,000
Albania	3,195,000
Algeria	35,423,000
Andorra	84,082
Angola	18,993,000
	↑ Leading digit

Chances are, the leading digit will be a 1 more often than a 2. And 2s would probably occur more often than 3s, and so on.

This odd phenomenon is Benford's Law. If a set of values were truly random, each leading digit would appear about 11% of the time, but Benford's Law predicts a logarithmic distribution. It occurs so regularly that it is even used in fraudulent accounting detection.

See the Wikipedia article for a more thorough discussion.

This is a simple experiment to see how many large, publicly accessible datasets satisfy Benford’s Law.

This site is on GitHub. Please help out by forking the project and adding more datasets.

Available Datasets

Twitter users by followers count

Leading digit frequency

132.62%▲
216.66%▲
311.8%▲
49.26%▲
57.63%▲
66.55%▲
75.76%▲
85.14%▲
94.56%▲

▲ Predicted by Benford's Law

Number of Records	38,670,514
Min Value	1
Max Value	4,706,631
Orders of Magnitude	6

Data Source:

http://www.infochimps.com/datasets/twitter-census-twitter-users-by-friends-count