Determining categories
IBM SPSS Statistics categorical variables are sometimes only partially coded. In other words, value labels might not exist for every data value in the .sav file. Because the MDM representation of data does not display values that do not have a corresponding MDM category, the SPSS Statistics SAV DSC uses a two-stage process to ensure that MDM categories exist for all data values in the .sav file:
▪First, it generates MDM categories for all labeled values.
▪By default, it then scans all cases in the .sav file and generates extra MDM categories for any data values that do not have a corresponding value label.
In general, you can obtain better results by editing the .
sav file and adding value labels for all data values in the file. A potential problem with generating categories from the case data is that the will assign MDM category values in the order in which unlabeled values are found in the .
sav file. If you update the .
sav file over time and you want the results of repeated analyses to be compatible, save the generated MDM document from the first analysis as a .mdd file (see
Creating an MDM document from an SPSS Statistics.sav file and use it to read the .
sav file for subsequent analyses.
You can change how the SPSS Statistics SAV DSC generates extra MDM categories by using the following settings (see
Properties and settings used by the SPSS Statistics SAV DSC):
|
To take this action
|
Use this setting
|
|
Scan only some cases to find extra categories, or do not perform any scanning.
|
ScanForCategories
|
|
Create an Other category for data values that do not have a corresponding MDM category.
|
AddOtherCategory
|
The UNICOM Intelligence Developer Documentation Library(DDL) includes a script that you can use to generate an .
mdd file without any category scanning or generation. For more information, see the entry for
CreateMddFromSavProperties.mrs in
Sample mrScriptBasic files.
See also