The inputs x1, x2, ... xn must be vectors of equal length with a data type of numeric, logical, char array, categorical, strings, datetime, duration, or cell array of character vectors. Missing values, such as NaN or NaT, are not counted by default. A NaN duration is missing too, whereas MATLAB counts it as a group of its own.

The following Name-Value pair is supported:

The label of the missing group is NaN for numeric inputs, <undefined> for categorical, <missing> for strings, {''} for cell arrays of character vectors, NaT for datetime, the text of a NaN element for duration, and the blank row itself for a char array.

As additional return values crosstab returns the chi-square statistics chisq, its p-value p and a cell array labels, containing the labels of each input argument.

With a single input, t is a column holding the count of each group, and chisq and p are NaN, there being no second variable to test independence against. A char row is a single observation.

A categorical input has a row or column for every one of its categories, an unused one included, holding zeros. The chi-square statistic and its p-value are computed over the rows and columns that hold observations, where MATLAB returns NaN for both whenever any row or column is empty. With the missing group counted, it follows every category here, whereas MATLAB returns a table one row longer than its labels when an unused category follows the last used one.

See also: grp2idx, tabulate

Source Code: crosstab