Missing Values

The SURVEYMEANS Procedure

Missing Values

When computing statistics for an analysis variable, PROC SURVEYMEANS omits observations with missing values for that variable. The procedure bases statistics for each variable only on observations that have nonmissing values for that variable. If you specify the MISSING option in the PROC SURVEYMEANS statement, the procedure treats missing values of a categorical variable as a valid category.

An observation is also excluded if it has a missing value for any STRATA or CLUSTER variable, unless the MISSING option is used.

If an observation has a missing value or a nonpositive value for the WEIGHT variable, then PROC SURVEYMEANS excludes that observation from the analysis.

The procedure performs univariate analysis and analyzes each VAR variable separately. Thus, the number of missing observations may be different for different variables. You can specify the keyword NMISS in the PROC SURVEYMEANS statement to display the number of missing values for each analysis variable in the "Statistics" table.

If you have missing values in your survey data for any reason (such as nonresponse), this can compromise the quality of your survey results. An observation without missing values is called a complete respondent, and an observation with missing values is called an incomplete respondent. If the complete respondents are different from the incomplete respondents with regard to a survey effect or outcome, then survey estimates will be biased and will not accurately represent the survey population. There are a variety of techniques in sample design and survey operations that can reduce nonresponse. Once data collection is complete, you can use imputation to replace missing values with acceptable values, and you can use sampling weight adjustments to compensate for nonresponse. You should complete this data preparation and adjustment before you analyze your data with PROC SURVEYMEANS. Refer to Cochran (1977), Kalton and Kaspyzyk (1986), and Brick and Kalton (1996) for more details.

If there is evidence indicating that complete respondents are different from incomplete respondents for your study, you can use the DOMAIN statement to compute the descriptive statistics "among complete respondents" from your survey data without imputation on incomplete respondents. See Example 13.3.

If missing values result in empty strata in the sample, then they will have an impact on the statistical computation, which uses the total number of strata. If all the observations in a stratum have missing weights or missing values for the current analysis variable, this stratum is an empty stratum . For example,

   data new;
      input stratum y z w; 
      datalines;
   1 . 13 40
   1 2  9  .
   1 .  5 25
   2 5 10 20
   2 8 60 15
   ;
   proc surveymeans df mean nobs nmiss; 
      strata stratum; 
      var y z;
      weight w;
   run;

You analyze variable Y and Z, with weight variable W and stratum variable STRATUM. For variable Y, all observations have missing values or missing weights in STRATUM=1, therefore, the analysis for variable Y uses only observations in STRATUM=2. Thus, for variable Y, STRATUM=1 is an empty stratum and STRATUM=2 is a non-empty stratum. Note, however, that STRATUM=1 is a non-empty stratum for variable Z.

If your sample design contains stratification, PROC SURVEYMEANS analyzes only the data in non-empty strata. Therefore, the total number of strata for an analysis variable means the total number of non-empty strata. In this example, the total number of strata for Y and Z is one and two, respectively.

Chapter Contents
Previous
Next
Top