IBM SPSS Statistics string variables whose width is more than eight bytes are known as long-string variables, and the SPSS Statistics SAV DSC always maps these to MDM text variables.

All other IBM SPSS Statistics string variables are known as short-string variables, and the SPSS Statistics SAV DSC interprets these in a similar way to numeric variables. By default, when generating an MDM document from the .sav file or accessing the case data with no metadata, the maps short-string variables that have at least one value label that is not a user-missing value to MDM single-response categorical variables. All other short-string variables are mapped to MDM text variables.

Value labels are case sensitive. To change how the determines if a short-string variable is categorical, see Determining categorical variables.