The statistics recorded for numeric variables are detailed in the section VAR Statement.

For classification variables, a *level* is a distinct observed value after formatting, removal of beginning and ending white space, and capitalization. For example,
the values `MyLevel`

and `MYLEVEL`

are treated as a single level in the data set. Classification variables can be numeric, and the same levelization rules apply.
For example, `3.000002`

and `3.0000001`

are treated as the same level if they are formatted using `BEST3`

.

Frequencies should be integers. If a noninteger frequency is specified, it is rounded to the nearest integer for calculations. Weights do not need to be integers. Negative frequencies and weights are treated as 0.