Troubleshooting
Problem
My data file consists of variables that are ratings from 1 to 10. I used SPSS Missing Value Analysis (MVA command) , with the regression method of imputation. to fill in missng attribute ratings. It is prediciting some values to be less than 1 or greater than 10, but there are no values in the file that would be out of the range. How can I constrain SPSS to fill in missing values with numbers between 1 and 10?
Resolving The Problem
The MVA command, which runs by the menu system from "Analyze>Missing Value Analysis", allows you to save a single sample with imputed values. The MVA procedure has no option to constrain the bounds of the imputed values or to round the imputed values. Constraints for imputed values can be set by the Multiple Imputation procedure, which was added to the Missing Values module in SPSS Statistics 17.0. See Technote 1477608 for a description of the Multiple Imputation procedure and its capabilities. Multiple Imputation is available from the menu system at
Analyze>Multiple Imputation>Impute Missing Data Values. Click the Constraints tab of the "Impute Missing Values" dialog, The lower and upper limits of the imputed values are entered in the Minimum and Maximum boxes, respectively, for each variable to be imputed. For the problem you described, you would enter 1 for Minimum and 10 for Maximum for each of the ratings variables. The level of rounding is controlled by entering a value in the Rounding box. By default, imputed values are not rounded. To round imputed values, specify a numeric multiplier value. Imputed values are rounded to the nearest integer multiple of the specified multiplier. For example, if you enter a 1 in the Rounding box, imputed values are rounded to the nearest integer; if you enter 0.1, imputed values are rounded to the nearest tenth. When running Multiple Imputation, these choices are specified in /CONSTRAINTS subcommands, as in:
MULTIPLE IMPUTATION salary time education
/IMPUTE METHOD=AUTO NIMPUTATIONS=5 MAXPCTMISSING=NONE MAXCASEDRAWS=50 MAXPARAMDRAWS=2
/CONSTRAINTS salary( MIN=0)
/CONSTRAINTS time( MIN=0 MAX=50)
/CONSTRAINTS education( MIN=0.0 MAX=21.0 RND=1.0)
/MISSINGSUMMARIES NONE
/IMPUTATIONSUMMARIES MODELS
/OUTFILE IMPUTATIONS=employ_impute .
In the above command, 5 samples with imputed data, along with the original, unimputed data, are stored in the data set employ_impute. A minimum value of 0 is imposed on imputed values for SALARY; a minimum of 0 and maximum of 50 for imputed values of TIME; a minimum of 0 and maximum of 21 for imputed values of EDUCATION. Imputed values for EDUCATION are rounded to the nearest integer.
Related Information
Historical Number
59672
Was this topic helpful?
Document Information
Modified date:
16 April 2020
UID
swg21478542