IBM SPSS Statistics uses the term system-missing for values assigned by the program when values in your data are undefined according to the format type you have specified, when a numeric field is blank, or when a value resulting from a transformation command is undefined. Numeric system-missing values are displayed as periods. Text variables cannot have system-missing values, because any character is legal in a string variable.

System-missing values are stored in an IBM SPSS Statistics variable as the largest possible negative number that can be stored in eight bytes (0xFFFFFFFF). If the UNICOM Intelligence Metadata Model (MDM) variable's value is NULL, the SPSS Statistics SAV DSC writes the system-missing value to the corresponding IBM SPSS Statistics variable. To write a different value to the IBM SPSS Statistics variable, use the NullValue property.

The SPSS Statistics SAV DSC exports an MDM variable's element as user-missing if the flUser flag is set on the element. The IBM SPSS Statistics category values for user-missing are the next category values available.

The UNICOM Intelligence authoring tools set the flUser flag on the following special categories: Don't Know, No Answer, and Refused. The SPSS Statistics SAV DSC exports these as user-missing values 1, 2, and 3. If there are more than three categories flagged as flUser, the fourth and subsequent categories are handled as regular categories, except that they will use the MissingValue property if you have specified it.