Skip to content


Outcomes and treatments are commonly described in contingency tables, which are at least 2 rows by 2 columns, but can be multiple rows and columns, hence the alternate term r x c table.

The term “contingency table” dates to 1904 and Karl Pearson’s paper “On Contingency and its Relation to Normal Correlation.” His 2 x 2 example, (from the original), correlated the presence or absence a smallpox vaccination mark (cicatrix) with disease outcomes in the 1890 smallpox epidemic. The column and row totals are termed marginal totals, or simply marginals, the grand total is in the lower right. The death rate among those with no vaccination is more than seven times the rate of those with a vaccination.

Pearson’s paper was published towards the end of a century-long effort to establish and promote vaccinations for smallpox, which afflicted two-thirds of the  children born in Europe in the 1700’s. The paper was the first fruit of a grant “in furtherance of research and higher work” in mathematics to the University of London from the “Worshipful Company of Drapers.

That, of course, is its short name; the full name is “The Master and Wardens and Brethren and Sisters of the Guild or Fraternity of the Blessed Mary the Virgin of the Mystery of Drapers of the City of London.”

A draper was a cloth merchant; the Drapers Company (actually a guild) got its royal charter in 1364. Queen Elizabeth II (the current queen of England) is a member. Today the Company exists primarily as a charitable and educational organization.’s Categorical Data Analysis course covers the use of logistic and Poisson regression to analyze contingency table data.