Data Model > Available DSCs > SPSS Statistics SAV DSC > Reading from an SPSS Statistics .sav file > Variable definitions when reading from a .sav file > IBM SPSS Statistics multiple response sets
 
IBM SPSS Statistics multiple response sets
The SPSS Statistics SAV DSC maps IBM SPSS Statistics multiple response sets (both multiple category sets and multiple dichotomy sets) that are based on numeric or string variables to MDM multiple-response categorical variables. However, the results might be unexpected in the following situations:
If more than one label list is used in an IBM SPSS Statistics multiple response set.
If the multiple response set consists of less than three variables.
If the multiple response set contains a mixture of numeric and string variables.
In addition, if the same IBM SPSS Statistics variable is used in more than one multiple response set, the SPSS Statistics SAV DSC attempts to create a single MDM categorical variable that represents the superset of the variables contained in the overlapping sets. This is successful only if the overlapping sets are all multiple category sets or all multiple dichotomy sets and, if dichotomous, they have the same counted value (that is, the value that indicates a "yes" response).
For example, in the 1991 U.S. General Social Survey example .sav file, there are nine health-related variables named hlth1 through hlth9. Suppose you create two multiple response sets as shown in the following table, making sure that the other options are the same for both multiple response sets. For example, choose either Categories or Dichotomies and the same Counted Value for both sets.
Name
Label
Variables
$ms1
Drug-related problems
hlth4, hlth5, and hlth8
$ms2
Child-related problems
hlth7 and hlth8
When the SPSS Statistics SAV DSC reads the .sav file, it creates one MDM multiple-response categorical variable called ms1_ms2, with a description of Drug-related problems - Child-related problems, and four categories based on the hlth4, hlth5, hlth7, and hlth8 variables. Note that the other health-related variables that were not included in either of the multiple response sets (hlth1, hlth2, hlth3, hlth6, and hlth9) become MDM single-response categorical variables. When preparing a .sav file for reading by the MDM, you might therefore want to create a multiple response set containing all of the variables that store the responses to a multiple response question.
When the response set contains system-missing values
If all the variables in a multiple dichotomy set contain IBM SPSS Statistics system-missing values, the SPSS Statistics SAV DSC sets the value of the corresponding MDM variable to a NULL value.
In contrast, if all the variables in a multiple category set contain IBM SPSS Statistics system-missing values, the SPSS Statistics SAV DSC sets the value of the corresponding MDM variable to an empty categorical value. This might cause problems if you tabulate the .sav file, because a case whose multiple category set variables are all set to the system-missing value will be included in the base calculation. To resolve this problem, use the CategorySet property (see Properties and settings used by the SPSS Statistics SAV DSC) when you create the .sav file to force the MDM variable to be exported as a multiple dichotomy set. Alternatively, in UNICOM Intelligence Reporter, define an expression for the base as described in "Restricting a Base Using an Expression" in the UNICOM Intelligence Reporter User's Guide.
The following example expression exclude rows whose variable myMPVariable contains an empty categorical response:
base('myMPVariable > {}')
See also
Variable definitions when reading from a .sav file